the relationship between sleep duration and working memory ... · performance in adults, the...
TRANSCRIPT
1
The relationship between
sleep duration and
working memory in children
Rebecca Ashton
D.Ed.Psy thesis Volume 1, 2016
University College London
2
Contents
Abstract .................................................................................................................................................. 6
Background ....................................................................................................................................... 6
Aims .................................................................................................................................................... 6
Sample ............................................................................................................................................... 6
Methods ............................................................................................................................................. 6
Results ............................................................................................................................................... 7
Conclusions ....................................................................................................................................... 7
Literature review ................................................................................................................................... 8
Introduction ........................................................................................................................................ 9
What is sleep? ................................................................................................................................ 10
Development of sleep .................................................................................................................... 11
Cultural influences on sleep .......................................................................................................... 13
Functions of sleep .......................................................................................................................... 16
Working memory ............................................................................................................................ 17
Development of working memory and its neural substrates .................................................... 20
Ways to improve working memory ............................................................................................... 22
Neural mechanisms by which sleep may affect working memory ........................................... 22
Sleep and working memory in children: a systematic literature review .................................. 25
Overview of associations............................................................................................................... 31
No relationship between sleep and working memory? ......................................................... 33
A positive linear relationship between sleep and working memory? .................................. 35
A negative linear relationship between sleep and working memory? ................................. 37
An inverted U relationship between sleep and working memory? ...................................... 39
Weight of evidence assessment .................................................................................................. 41
Conclusions ..................................................................................................................................... 44
Future research .............................................................................................................................. 46
Empirical study ................................................................................................................................... 48
Introduction ...................................................................................................................................... 49
Method ............................................................................................................................................. 55
Study design ............................................................................................................................... 55
Measures ..................................................................................................................................... 57
Intervention .................................................................................................................................. 61
Participants .................................................................................................................................. 63
Results ............................................................................................................................................. 72
Statistical analysis ...................................................................................................................... 72
Baseline measures ..................................................................................................................... 73
Measurement issues .................................................................................................................. 80
Intervention effects ..................................................................................................................... 84
Programme acceptability ........................................................................................................... 94
3
Discussion ...................................................................................................................................... 98
Key findings ................................................................................................................................ 98
Key methodological issues ..................................................................................................... 101
Key theoretical issues ............................................................................................................. 105
Future research ........................................................................................................................ 107
Critical appraisal .............................................................................................................................. 110
The process of research development ..................................................................................... 111
Research questions, methodology and epistemology ........................................................... 113
Intervention-led research ............................................................................................................ 117
Risky research ............................................................................................................................. 119
Using advice from supervisors .................................................................................................. 122
Literature review ...................................................................................................................... 122
Dependence on others ........................................................................................................... 124
Data input and analysis .......................................................................................................... 125
The (un)importance of sleep? .................................................................................................... 126
A distinct contribution .................................................................................................................. 128
Methodology ............................................................................................................................. 128
The relationship between sleep and working memory in children.................................... 129
The future for sleep interventions .......................................................................................... 130
The future for sleep research ................................................................................................. 133
Key messages for educational psychologists ..................................................................... 135
Appendix 1: Ethics approval ......................................................................................................... 163
Appendix 2: Project date planner .................................................................................................. 165
Appendix 3: Sleep diary blank form .............................................................................................. 166
Appendix 4: Actiwatch wearer’s guide .......................................................................................... 167
Appendix 5: Sleep knowledge questionnaire .............................................................................. 169
Appendix 6: Amendments to ACES materials............................................................................. 170
Appendix 7: Flyer to schools ......................................................................................................... 175
Appendix 8: Teacher information & consent form ...................................................................... 176
Appendix 9: Parent and child information & consent form ........................................................ 180
Appendix 10: Scattergrams showing relationships between WM and sleep variables ......... 182
4
Figures and tables
Figure 1: Model of working memory components (Baddeley, 2000, p.421) .............................. 18
Figure 2: Key neuroanatomical areas (retrieved from brightfocus.org on 12 August 2015) ... 21
Figure 3: Brain areas involved in children’s working memory (retrieved from
gettyimages.co.uk on 12 August 2015) ........................................................................................... 23
Figure 4: Literature search strategy flow diagram ......................................................................... 27
Figure 5: Flow diagram of participation throughout the study ...................................................... 66
Figure 6: Sleep knowledge at Times 1, 2 and 3 ............................................................................ 86
Figure 7: Mean total time asleep (in minutes) as measured by actigraphy at Times 1, 2 and 3
.............................................................................................................................................................. 89
Figure 8: Mean sleep efficiency as measured by actigraphy at Times 1, 2 and 3 .................... 89
Figure 9: Verbal WM raw scores at Times 1, 2 and 3 ................................................................... 91
Figure 10: Visual WM raw scores at Times 1, 2 and 3 ................................................................. 91
Figure 11: From research question to epistemology ................................................................... 116
Table 1: Research relating measures of sleep duration and working memory in children ...... 28
Table 2: Results of selected studies ................................................................................................ 32
Table 3: Weight of evidence analysis .............................................................................................. 43
Table 4: IDACI scores with corresponding ranks and deciles ..................................................... 68
Table 5: Participant demographics .................................................................................................. 68
Table 6: Numbers of sleep diaries completed ................................................................................ 69
Table 7: Numbers of Actiwatches used .......................................................................................... 70
Table 8: Numbers of children with usable Actiwatch data ............................................................ 71
Table 9: Sleep duration at baseline ................................................................................................. 74
Table 10: Baseline working memory scores .................................................................................. 76
Table 11: Tests of normality for working memory and diary data at Time 1.............................. 77
Table 12: Correlations between sleep diary and working memory variables at Time 1 .......... 78
Table 13: Correlations between sleep actigraphy and WM variables at Time 1 ....................... 79
Table 14: Descriptive statistics for sleep diary and actigraphy data at Time 1 ......................... 80
Table 15: Correlations between diary and actigraphy measures at Time 1 .............................. 81
Table 16: Times of falling asleep and waking up at Time 1 ......................................................... 82
Table 17: Responses to sleep knowledge questionnaire at Time 1 ........................................... 84
Table 18: Effects of intervention on sleep knowledge .................................................................. 85
Table 19: Descriptive statistics for actigraphy data at Times 1, 2 and 3 .................................... 87
Table 20: Tests of normality for change scores in sleep diary and WM variables ................... 92
Table 21: Mann-Whitney U tests of changes in WM scores between control and intervention
groups .................................................................................................................................................. 93
Table 22: Correlations between changes in total time asleep and changes in WM scores .... 94
Table 23: Feedback from children about the intervention programme ....................................... 96
Table 24: Realised risks and their management ......................................................................... 120
5
Acknowledgements
With thanks to all the children who participated in this study, and their teachers:
Beverley Asgharyzadegar, Blackburn The Redeemer CE Primary School
Rosie Barnes, Blackburn The Redeemer CE Primary School
Stewart Crewe, St Mary & St Joseph RC Primary School
Dawn Furnell, St Barnabas CE Primary School
Jane Galloway, St Aidan’s CE Primary School
Shahida Khan, St Antony’s RC Primary School
Sara Parkinson, Holy Souls RC Primary School
Ajanta Wilkinson, Holy Trinity RC Primary School
Thanks also to Charlotte Smart, honorary assistant psychologist, without whom the
data collection for this research would not have been possible.
6
Abstract
Background
While sleep appears to have a robust relationship with working memory (WM)
performance in adults, the picture in children is less clear.
Aims
The current study investigated the relationship between sleep duration and WM in
children, with both correlational and experimental aspects to the design. The
research also aims to clarify valid and effective ways to measure sleep duration with
children in their home setting.
Sample
The participants were eight classes of children aged 9-10 years. A total of 220
children were involved, of which 107 experienced a sleep education intervention and
113 formed a control group.
Methods
Data were collected via sleep diaries, sleep knowledge questionnaires and tests of
WM and attention. In each class, 10 children also wore an actigraph to record
7
objective sleep data. The intervention was adapted from the ACES programme
(Blunden, 2007b).
Results
Most sleep variables showed no significant correlation with WM scores at baseline.
There was a negative association between diary sleep duration and verbal WM,
which was not significant after controlling for socioeconomic status. Actigraphy
showed no relationship between sleep duration and WM, but there was a positive
correlation between sleep efficiency and visual WM when socioeconomic status was
partialled out. The intervention had a sustained effect on sleep knowledge but not
on sleep duration or efficiency. With regard to methodology, sleep diary data were
not systematically affected by introducing actigraphy as an objective sleep measure.
Repeated measurements proved difficult with this age group and there were higher
dropout rates from sleep diary completion than from repeatedly wearing an
actigraph.
Conclusions
The relationship between sleep and WM in 9-10 year old children is weak at best, but
suggests that shorter sleep with higher efficiency is related to better WM scores. Sleep
duration could be effectively measured by actigraphy alone, without sleep diaries.
Sleep interventions may be more effective if targeted rather than universal, and future
research could use a neurodevelopmental perspective to help explain why the
relationship between sleep and WM appears to change with maturity.
8
Literature review
9
Introduction
Sleep and working memory are two areas which have significant impact on
outcomes for children. Sleep quality and duration can affect a range of other
variables including emotional regulation (Gruber, Cassoff, Frenette, Wiebe, &
Carrier, 2012), executive functioning (Sadeh, Gruber, & Raviv, 2002), academic
attainment (Wolfson & Carskadon, 1998) and physical health such as obesity (Chen,
Beydoun, & Wang, 2008). Working memory has been shown to predict academic
attainment (Alloway & Alloway, 2010), especially in Maths (Gathercole, Pickering,
Knight, & Stegmann, 2004). Working memory is even thought to be a core cognitive
problem in psychiatric conditions including schizophrenia (Lee & Park, 2005) and
attention deficit hyperactivity disorder (Martinussen, Hayden, Hogg-Johnson, &
Tannock, 2005).
There may also be a relationship between sleep and working memory. In a review,
Kopasz et al. (2010) concluded that, “most studies support the hypothesis that sleep
facilitates working memory as well as memory consolidation in children and
adolescents” (p.167). However, they reviewed only three studies including
measures of working memory (as opposed to measures of short term memory,
learning or implicit memory), with only one of those showing a relationship between
working memory and sleep variables (Steenari et al., 2003).
It is possible to intervene directly to improve working memory, at least in the short
term, with adaptive programmes which train capacity over hundreds of trials (Melby-
Lervag & Hulme, 2013; Shipstead, Redick, & Engle, 2012). There is a large and
increasing body of evidence in this field, with controversy over how effective such
training can be in the long term and how generalizable the effects might be
(Gathercole, Dunning, & Holmes, 2012; Redick et al., 2013; Hulme & Melby-Lervag,
10
2012). There are also direct interventions under development to improve sleep
knowledge and sleep behaviours, although these are at a much earlier stage of
validation (Blunden, Chapman, & Rigney, 2012).
This paper aims to synthesise knowledge from the two applied fields of working
memory and sleep behaviours. The purpose is to evaluate current understandings
of how the two might be linked and possible next steps for research.
What is sleep?
Sleeping is a curious phenomenon, in which animals enter a, “reversible behavioural
state of perceptual disengagement from and unresponsiveness to the environment”
(Carskadon & Dement, 2011, p. 16). Sleep is a universal behaviour found in all
animals, follows a regulated pattern in each species and there are deleterious
effects if this pattern is seriously disrupted (Cirelli & Tononi, 2008).
Although we tend to refer to being either, “asleep,” or, “awake”, humans experience
a range of states of consciousness at different times (Tassi & Muzet, 2001). The
sleep state can be broadly divided into two main types: rapid eye-movement (REM)
and non rapid eye-movement (NREM), and in humans NREM can be further divided
into four stages. Across a period of sleep, brain activity cycles through the stages of
REM and NREM sleep which are distinguishable by the frequency of neuronal
electrical activity (Walker & Stickgold, 2006). The pattern of sleep stages is known
as sleep architecture. Healthy adults usually progress through the NREM stages
and back, followed by a period of REM sleep. This cycle takes around 90 minutes
and is then repeated, with increasing REM durations towards the end of the sleep
period (Carskadon & Dement, 2011).
11
Measurements of sleep include sleep duration (total amount of time spent asleep in
a 24-hour period) and sleep quality (subjectively feeling refreshed from sleep and
objectively showing sleep efficiency, i.e. spending high proportions of the time in bed
actually asleep). There are a number of ways to measure these aspects of sleep.
Subjective measures include sleep diaries, usually recording bed times, sleep times,
wake times and sleepiness over the course of one week, and questionnaires about
usual sleep habits which do not capture daily variations. Objective measures
include polysomnography, in which the participant wears recording electrodes
overnight, and actigraphy which involves a watch-like device with an accelerometer
to measure movement, usually worn on the non-dominant wrist. Polysomnography
is done in a sleep laboratory, usually for one or two nights, and can be combined
with video recordings to match up observable behaviours with electrical discharge
patterns from the brain. Actigraphy can be carried out at home, and is usually done
over the course of a week, followed by data analysis using software which infers
periods of sleep or wakefulness from the intensity and duration of movements
recorded.
Development of sleep
Sleep duration and architecture changes over the course of maturation. Total sleep
duration within a 24-hour period decreases, from around 15 hours for babies to nine
hours in middle childhood and less than eight hours by the age of 16 years (Galland,
Taylor, Elder, & Herbison, 2012; Gradisar, Gardner, & Dohnt, 2011; Iglowstein,
Jenni, Molinari, & Largo, 2003).
According to Crabtree and Williams (2009), newborn babies sleep little and often,
12
starting with active sleep (a precursor to REM sleep in which the baby moves often
because the neural mechanisms that govern muscle paralysis during sleep are not
yet developed) and moving into quiet sleep. By three months of age, most babies
are starting to sleep more at night than in the day, and show definable stages of
NREM during quiet sleep. At six months, sleep episodes usually start with NREM
sleep and move into REM sleep later in the cycle (similar to the normal adult
pattern). The ratio of REM to NREM sleep changes, from more REM sleep in
babies to more NREM sleep by the end of the first year of life.
After the first year, napping in the daytime tends to decrease (although this pattern
is highly dependent on the context) so that almost all sleep occurs at night-time by
the age of four years. Night-time wakings are common in the pre-school years
(Schwichtenberg & Goodlin-Jones, 2010). In the primary school years, sleep habits
become less variable with most children getting 8-10 hours per night (Galland et al.,
2012). Girls tend to sleep longer than boys, and both genders report feeling more
sleepy in the daytime as they progress through primary school (Sadeh, Ravi, &
Gruber, 2000).
Adolescence marks a period of rapid change in many aspects of sleep. Some of
these changes are maturational and relate to the hormonal changes of puberty,
such as the circadian shift towards later bed times and later wake times, while
others are age-related, such as the overall reduction in total sleep time accounted
for by reductions in NREM sleep time (Colrain & Baker, 2011). Over the course of
the school years, the discrepancy between school nights and weekends increases
(Gradisar et al., 2011), suggesting that young people may be getting insufficient
sleep when a wake-time is externally set and then recovering when the schedule
allows.
13
The focus of this research was children towards the end of primary school (age 9-10
years), at which point most children have established a sleeping pattern without
daytime naps (Thorleifsdottir, Bjarnsson, Benediktsdottir, Gislason, &
Kristbjarnarson, 2002) and are starting to take more responsibility for their own
behaviours (e.g. Such & Walker, 2004), while not yet undergoing the rapid changes
of puberty (e.g. Laberge et al., 2001).
Cultural influences on sleep
Sleep patterns, in adults and during development, are influenced not only by biology
but also by culture (Jenni & O'Connor, 2005). Community norms and expectations
may have an important role to play in when, where and for how long people sleep.
For example, in China children tend to have a shorter night-time sleep duration than
those living in the USA (Liu, Liu, & Wang, 2003; Liu, Liu, Owens, & Kaplan, 2005).
The authors suggest that some reasons for this difference include earlier school
start times in China, greater homework demands, napping during the daytime and
increased prevalence of bed-sharing. The frequency of sleep problems as reported
by parents is also higher in China than the USA in these studies, which may be
partly due to greater parental awareness of what happens during the night if the
parent is sharing the room or even the bed with their child.
Recommendations for sleep practices are therefore likely to reflect what we know
about biological needs but also culturally normative behaviours. Sleep hygiene is
often used to describe a person’s bedtime environment and routine, with good sleep
hygiene including behaviours which promote sleep and excluding behaviours which
might make it harder to get to sleep or stay asleep (e.g. Mastin, Bryson, & Corwyn,
2006). The fit between an individual’s sleep need and what is considered
appropriate sleep hygiene in their context may be a key factor in whether a sleep
14
problem is perceived. For example, in Italy children may often be invited to stay up
late with the family. Falling asleep during a family gathering is considered perfectly
acceptable, rather than being expected to go to sleep in their bedroom and to go to
bed earlier than older members of the family (Ottaviano, Giannotti, Cortesi, Bruni, &
Ottaviano, 1996). Interestingly, Italian adolescents report themselves to have more
regular bedtime routines than American teenagers, so the difference in sleep
hygiene practices earlier in childhood appears to develop in opposite directions in
these two countries (LeBourgeois, Giannotti, Cortesi, Wolfson, & Harsh, 2005).
Again, this difference may reflect cultural factors: perhaps Italian adolescents are
expected to develop their own sleep habits and to be responsible for the
consequences of those habits, while American children are expected to follow adult-
set rules which they rebel against during their teenage years.
As well as broad cultural norms such as these national approaches, individual
families may have their own cultural practices around sleep which affect a child’s
sleep behaviours. One obvious difference between families might be whether the
child usually sleeps in their own room, or shares the bedroom with another family
member. One might hypothesise that room sharing would increase the probability of
sleep problems, due to the other person creating stimuli which disturb sleep,
however the evidence suggests that sharing a room with a sibling makes no
significant difference to children’s sleep behaviours (Kahn et al., 1989; Stein,
Mendelsohn, Obermeyer, Amromin, & Benca, 2001; Worthman & Brown, 2007).
Another family-level difference might be whether the child stays in the same home
throughout the week, or lives across more than one home (for example if the
parents are separated and have shared parenting arrangements). Likewise, family
stress would be likely to impact upon sleep behaviours (for example, witnessing
parents arguing or domestic abuse). Family members on different diurnal schedules
15
may also impact the sleep patterns of a child (for example, a parent who works
shifts or an older sibling who goes out until late at night). Surprisingly little research
has been published investigating the relationship of such living environments with
children’s sleep behaviours; in fact no evidence about the impact on sleep of living
in more than one home could be found. For shift-working parents, much has been
written about the adults feeling sleep deprived (e.g. Geiger-Brown, Trinkoff, &
Rogers, 2011) but nothing about the impact of their comings and goings on
children’s sleep. In terms of family stress, Meltzer and Mindell (2007) found
significant correlations between mothers’ reports of their own distress and of their
children’s sleep problems, although this finding could be an artefact of participants’
response bias.
The direction of cause and effect in sleep behaviours and sleeping context is not
always clear. For example, bed sharing and room sharing with parents was
correlated with sleep problems in Chinese children of school age (Li et al., 2008),
but this does not necessarily mean that sharing the sleeping environment caused
the sleep problems. It may be that having another person in the room generated
stimuli which disturbed the child’s sleep pattern, thus causing problems such as
night-time wakings and parasomnias (nightmares, sleep walking and other
disturbances during sleep). Alternatively, the child may have been experiencing
sleep problems so the adult decided to sleep with them in the hope of alleviating the
difficulties. A third possibility is that co-sleeping enabled the adult to be more aware
of the child’s sleep problems, although this explanation would not account for the
increased prevalence of problems outside of the bedroom such as daytime
sleepiness and bedtime resistance.
Finally, an increasingly prevalent cultural factor influencing sleep behaviours must
be the use of media. A full discussion of how television, computer, games console,
16
tablet and mobile phone usage might relate to sleep is beyond the remit of this
paper. However, a recent systematic review of this topic by Hale and Guan (2015)
suggests that there is a positive correlation between screen time and shorter sleep
duration in children of school age. The mechanisms by which media use might
affect sleep include elevated light levels, elevated arousal levels, motivation to sleep
or to stay awake, feelings of worry and feelings of anticipation. While it seems
obvious that increased media use may delay or disrupt sleep, this assumption is
being challenged by studies such as that of Tavernier and Willoughby (2014),
indicating that in adults at least, increased media use may be a way of coping with
sleep problems rather than causing them.
Functions of sleep
Given the risks involved in sleep, during which the individual is not able to respond
to threats, presumably it must fulfil important functions. It has been remarkably
difficult, however, to specify exactly why we sleep and whether the functions of
sleep are the same in all animals.
A common approach to researching sleep functions has been to monitor the results
of sleep deprivation. Paradigms include acute sleep deprivation (asking participants
to stay awake for a set period of time), longer-term sleep restriction (asking
participants to reduce their usual sleep duration for several nights) and recruiting
participants who are chronically sleep deprived (e.g. people with insomnia). A wide
range of functions have been shown to be impaired by sleep deprivation, including:
Metabolic and endocrine regulation, which can lead to weight gain (Gozal &
Kheirandish-Gozal, 2012)
17
Emotional regulation, towards both positive and negative stimuli (Simola,
Liukkonen, Pitkaranta, Pirinen, & Aronen, 2014)
Memory consolidation (Henderson, Weighall, Brown, & Gareth Gaskell,
2012)
Maintaining attention or vigilance (Lim & Dinges, 2008)
Other executive functions – although it is unclear how far decrements in
these tasks are related to impaired attention (Killgore, 2010)
Different aspects of sleep may be involved with different functions. For example,
NREM sleep and REM sleep may have complementary effects on long-term
memory formation, being involved in reorganisation and consolidation respectively
(Diekelmann & Born, 2010). In terms of emotional regulation, REM sleep appears to
be more important than NREM sleep (Walker & van der Helm, 2009).
Working memory
One function frequently found to be affected by sleep deprivation in adults is working
memory (Fortier-Brochu, Beaulieu-Bonneau, Ivers, & Morin, 2012; Lim & Dinges,
2010). Working memory is described by Alan Baddeley (2012) as, “a hypothetical
limited capacity system that provides the temporary storage and manipulation of
information that is necessary for performing a wide range of cognitive activities” (p.
7). Baddeley goes on to present the development of the model of working memory
which he set out and built upon with Graham Hitch, which includes the original
components of central executive, auditory and visuo-spatial elements along with an
episodic buffer for information integrated from different senses.
18
Figure 1: Model of working memory components (Baddeley, 2000, p.421)
Consistent with Baddeley and Hitch’s model, Alloway, Gathercole and Pickering
(2006) showed that short term memory (STM) is separate from working memory
(WM). Their analysis suggests that the phonological loop and visuo-spatial sketch
pad can act as simple storage mechanisms (hence separable verbal and spatial
STM), and when the task requires manipulation of the stored information, the central
executive becomes more involved (hence verbal and spatial WM are more highly
correlated than verbal and spatial STM because the same central executive
resource is used for WM but not STM tasks).
Long-term memory (LTM) is also distinguishable from WM, although WM may draw
upon LTM especially in chunking or categorising information based on existing
knowledge (Baddeley, 2012). The distinction between WM and LTM goes back to
19
William James’ 19th century description of primary and secondary memory, as
Jeneson & Squire (2012) remind us. Qualitatively, Alloway and Gathercole (2006)
describe the difference between WM and LTM being the, “catastrophic loss,” (p.
134) of information from WM when attention is taken away from it, or when capacity
is exceeded. Information in LTM, on the other hand, may be retrieved again in the
future. Some authors have suggested that all memory is LTM, with attentional
mechanisms bringing memories into short-term activation (Cowan & Chen, 2009).
Most researchers do posit that LTM is separate from WM, although there is
conflicting evidence about the neural substrates for each system (e.g. Jeneson &
Squire, 2012; Nee & Jonides, 2011).
There is ongoing controversy about the relationship between working memory and
attentional control. Baddeley (2000; 2012) conceptualises the central executive as
an attentional system, which interacts with the three “storage” elements to form
working memory (see Figure 1). Others see working memory as nothing more than
attentional control (Chun, 2011), although this view is not consistent with evidence
that WM measures can predict variance in IQ over and above attention measures
(Unsworth & Spillers, 2010). Overall it would appear WM is not a unitary construct,
and it may be an artefact of the combination of attention, STM and LTM or may be a
function in itself.
Whatever the nature of working memory, measures designed to tap this construct
have been shown to predict outcome variables including general intelligence (Suss,
Oberauer, Wittmann, Wilhelm, & Schulze, 2002; Conway, Cowan, Bunting,
Therriault, & Minkoff, 2002; Conway, Kane, & Engle, 2003; Ackerman, Beier, &
Boyle, 2005) and academic attainments (Bull & Scerif, 2001; Bull, Espy, & Wiebe,
2008; Cain, Oakhill, & Bryant, 2004; Gathercole et al., 2004).
20
Development of working memory and its neural substrates
Like many other cognitive skills, performance on tasks designed to measure working
memory improves as a child matures, continuing to improve into young adulthood
(McAuley & White, 2011). In the normal course of development, simpler tasks can
be performed at adult levels by an earlier age (e.g. delayed face recognition is at
ceiling by nine years old), while more complex tasks such as organising strategies to
aid WM continue to show improvement across childhood (Luciana, Conklin, Hooper,
& Yarger, 2005).
In terms of the neural basis of working memory, a meta-analysis of imaging studies
involving adults identifies a fronto-parietal network which seems to be consistently
involved in any task making WM demands (Rottschy et al., 2012), and longitudinal
data link the development of both the frontal and parietal lobes (see Figure 2) to
improvements in working memory performance (Darki & Klingberg, 2015; Tamnes et
al., 2013).
21
Figure 2: Key neuroanatomical areas (retrieved from brightfocus.org on 12 August 2015)
There is also some evidence that the neural substrates recruited for working
memory tasks shift as children mature. Finn, Sheridan, Kam, Hinshaw and
D’Esposito (2010) found that young women aged 15 used the hippocampus (see
Figure 2) as well as the fronto-parietal network for WM tasks, but by the age of 18
this hippocampal activation was no longer significant. In 9-10 year old children,
Chaddock et al. (2010) found a correlation between hippocampal volume and a task
designed to measure relational memory, which was very similar to other reported
measures of visual working memory. Thomason et al. (2009) also found that
children aged 7-12 used parahippocampal areas for WM tasks more than adults,
and for spatial WM tasks the occipital cortex was also more active in children than in
adults.
This change in how the brain handles working memory tasks corresponds with
22
maturation processes, with prefrontal cortex being the last area to undergo synaptic
pruning and myelination of white matter (Lenroot & Giedd, 2010) and therefore not
ready to play a greater part in WM processing until later in development.
Ways to improve working memory
Because working memory seems to be an important contributor to achievement in a
range of areas, much effort has been put into finding ways to improve WM capacity.
Computerised programmes such as Cogmed (e.g. Klingberg, Forssberg, &
Westerberg, 2002), Memory Booster (St Clair-Thompson, Stevens, Hunt, & Bolder,
2010) and Jungle Memory (e.g. Alloway, 2012) have been commercially available
for some years now. Morrison & Chein’s (2011) review also included strategy
training approaches, although the results have been less encouraging in this area.
In some cases, commercial packages have been launched with no published
evidence of their effectiveness (e.g. Memory Magic, reviewed by Tayler, 2015).
Looking at working memory as part of a broader group of executive functions,
Diamond & Lee’s (2011) wide ranging review included potential interventions such
as martial arts, high-intensity exercise, yoga and mindfulness. Food intake can also
affect working memory performance in children (Wesnes, Pincock, Richardson,
Helm, & Hails, 2003). However, to date no research is available which looks at
sleep as a target for intervention which may improve working memory.
Neural mechanisms by which sleep may affect working memory
In adults, imaging studies have confirmed that sleep deprivation affects working
memory performance by changing activation in prefrontal and parietal areas (see
Figure 3), with the nature of the changes being dependent on task difficulty (Chee &
23
Choo, 2004; Choo, Lee, Venkatraman, Sheu, & Chee, 2005; Lythe, Williams,
Anderson, Libri, & Mehta, 2012; Wu et al., 2006). Recent data suggest that sleep
deprivation may specifically decrease connectivity in the prefrontal cortex (Verweij et
al., 2014), although the molecular mechanisms by which sleep affects frontoparietal
functioning remain unclear.
Figure 3: Brain areas involved in children’s working memory (retrieved from gettyimages.co.uk on 12 August 2015)
Experimental studies in rodents have focused on the hippocampus (see Figure 3),
which is thought to be particularly sensitive to sleep deprivation in both rodents
(Ruskin, Liu, Dunn, Bazan, & LaHoste, 2004) and humans (Van Der Werf et al.,
2009). Such research provides initial indications that sleep plays a role in protein
synthesis, and sleep deprivation can lead to changes in the expression of genes that
govern this process (Hagewoud et al., 2010; Vecsey et al., 2012). In addition,
rodent studies have shown that sleep deprivation can inhibit the genesis of new
neurons in the hippocampus (Hairston et al., 2005; Ruben et al., 2003), and that this
24
process can be reversed by growth hormone (Kim, Grover, Bertolotti, & Green,
2010). In humans, it would seem that the growth hormone usually available to
normally developing children is not sufficient to prevent sleep deprivation from
inhibiting neuronal growth, as Taki et al. (2012) report that sleep duration is
correlated with hippocampal volume in children.
Putting the evidence regarding the hippocampus together, it is theoretically plausible
that sleep deprivation could cause slower neurogenesis in this brain region, and
because children use the hippocampus more than adults for working memory tasks,
this could be a mechanism by which reduced sleep impairs WM performance in
children. Alternatively, or perhaps in addition, sleep deprivation may impede the
development of the fronto-parietal network which subserves mature working
memory processes. Consistent with this second hypothesis, adolescents reporting
chronically poor sleep showed less activation in the frontal cortex during a cognitive
control task (Telzer, Fuligni, Lieberman, & Galvín, 2013).
25
Sleep and working memory in children: a systematic literature review
The key question for this review was, “what are the effects of sleep duration on
working memory in children?” The age range 9-10 years was of particular interest
(see section on development of sleep) but all ages were considered and then
studies with children under 16 years old were selected. A literature search was
conducted on 1st November 2014 using PsychInfo and PubMed. The terms “sleep”
and “working memory” were combined, resulting in 282 hits from PsychInfo and 360
from Pubmed. After excluding 181 duplicates, the total number of unique hits was
461.
These results were then filtered as follows:
Peer reviewed journal articles only
English language only
Excluding populations with conditions that may affect neuropsychological
functioning (e.g. studies of people with epilepsy, sleep apnoea, drug users
and elderly people)
Excluding empirical studies where sleep duration or working memory was
not a variable
This process resulted in 107 different articles. Two of these articles were meta-
analyses and formed the starting point for this review, since systematic reviews of
this nature are considered to be at the top of the hierarchy of evidence (Scott, Shaw,
& Joughin, 2001). Both meta-analyses are of studies involving adults aged at least
18 years, with one focusing on the cognitive effects of short-term sleep deprivation
(Lim & Dinges, 2010) and the other on long-term insomnia (Fortier-Brochu et al.,
2012). Both reviews found that working memory was one of the two most affected
cognitive functions (moderate combined effect sizes of r = -0.555 and r = -0.42
26
respectively). A further meta-analysis was identified outside the original search,
which concerned children aged 5-12 years (Astill, Van der Heijden, Van IJzendoorn,
& Van Someren, 2012). Working memory was not analysed as an individual
variable, but WM measures were categorised as executive functioning along with
response inhibition, planning, set-shifting and creative thinking. All the effect sizes
in this study were much smaller than in the adult meta-analyses, however due to the
large numbers of participants the association between sleep duration and executive
functioning (including WM) was statistically significant (14 studies with a total of
2390 participants; r = 0.07, p > 0.05). Taken together, these meta-analytic findings
suggested that there may be a link between sleep and working memory which
warranted further investigation.
Of the 107 articles returned, 17 included children in their remit. Three of these were
review papers and the remainder reported original research findings, of which 10
studies included measures of both sleep parameters (including duration, not just
EEG or ratings of sleep problems) and working memory (not just short term memory
such as forward digit span tasks). A further five studies researching the links
between working memory and sleep in children were found through wider searches
(replacing “working memory” with “cognition”, “IQ” and “functioning”, also searching
for these terms in Google Scholar and following up references listed in the selected
articles). Figure 4 represents the literature search strategy; Table 1 summarises the
15 selected studies resulting from this search.
27
Figure 4: Literature search strategy flow diagram
PubMed & PsychInfo search terms: “sleep” + “working memory”
461 articles
Filtered for peer reviewed, English language articles investigating sleep and working memory in ordinary populations
107 articles
Filtered for reports of original research including measures of sleep duration and working memory in children (0-16 years)
10 articles
Additional studies from wider searches meeting all criteria
15 articles
28
Table 1: Research relating measures of sleep duration and working memory in children
Authors & year
Participants Design Sleep measures WM measures Main results
Beebe, Difrancesco, Tlustos, McNally, & Holland (2009)
N = 6 Ages 13-16 years Community sample USA
Experimental sleep restriction
Actigraphy Sleep diary (self)
Computerised n-back task (0 and 2-back used)
No statistically significant association between WM performance after usual sleep and after sleep restriction. fMRI data showed that brain activity was magnified after the sleep restriction, suggesting that more mental effort was required to achieve similar performance.
Buckhalt, El-Sheikh, & Keller (2007) *
N = 166 Ages 8-9 years Community sample USA
Correlational study
Actigraphy Sleep diary (parent) Parent report of sleep problems
Woodcock-Johnson III (Numbers Reversed and Auditory WM tasks)
No association between any sleep parameter (sleep duration, efficiency, reported sleep problems) and WM. However, in low SES children, sleep efficiency predicted WM.
Calhoun et al. (2012)
N = 508 Ages 6-12 years Community sample USA
Correlational study
PSG Parent report of excessive daytime sleepiness
WISC-III Digit span PSG and parent report were unrelated. Parent report of sleepiness was significantly correlated with WM but measures of sleep duration and efficiency were not.
Cho et al. (2015) *
N = 1740 Ages 6-7 years Community sample Australia
Correlational study
Parent report of sleep duration, regularity and problems
AWMA Backwards digits and Mister X
Significant associations between sleep variables and verbal WM but not visual WM.
Geiger, Achermann, & Jenni (2010) *
N = 60 Ages 7-11 years Community sample Switzerland
Correlational study
CCQT (parent report) Actigraphy
WISC-IV Working memory index
Questionnaire and actigraphy sleep duration was negatively correlated with WM; significant for weekend data and approaching significance for weekdays.
29
Authors & year
Participants Design Sleep measures WM measures Main results
Gradisar, Terrill, Johnston, & Douglas (2008)
N = 143 Ages 13-18 Community sample Australia
Correlational study
Online sleep questionnaire (self)
Letter-number sequencing adapted from WISC-IV; operation span task
Inverted-U relationship: WM best for participants with medium sleep duration, compared to those with shorter and longer sleep durations.
Gruber et al. (2010) *
N = 39 Ages 7-11 years Community sample Canada
Correlational study
Actigraphy MSLT Sleep diary (parent) Subjective report of sleepiness
WISC-IV Digit span & Letter-Number Sequencing
No significant associations between sleep duration and WM. Associations between sleep efficiency and WM not reported.
Jiang et al. (2011) N = 40 Ages 13-20 years 20 aged 13-16 and 20 aged 18-20 Community sample China
Experimental sleep restriction
Actigraphy Sleep diary (self)
Computerised task of letter manipulation and matching, also arithmetic task
Association found between sleep restriction and poorer arithmetic scores in the 13-16 age group.
Konen, Dirk, & Schmiedek (2015)
N = 110 Ages 8-11 years Community sample Germany
Repeated measures (3 times per day for 31 days), no experimental manipulations
Smartphone sleep diary (self)
Smartphone verbal and visual updating tasks
Inverted-U relationship: WM morning performance best after average sleep duration but worse after extended or restricted nights’ sleep.
Sadeh, Gruber, & Raviv (2003) *
N = 77 Ages 9-12 years Community sample Israel
Experimental sleep restriction or extension
Actigraphy Computerised visual forward and backward digit span
Sleep extension significantly enhanced forward (but not backward) span.
30
Authors & year
Participants Design Sleep measures WM measures Main results
Steenari et al. (2003)
N = 60 Ages 6-13 years Community sample Finland
Correlational study
Actigraphy Sleep diary (parent)
Computerised verbal and visual n-back tasks (0, 1 and 2-back used)
Sleep duration and efficiency were significantly correlated with error rate in the more demanding WM tasks
Van der Heijden, de Sonneville, & Swaab (2013)
N = 303 Ages 7-12 years Community sample Netherlands
Correlational study
Sleep diary (parent)
Amsterdam Neuropsychological Tasks: 4 Letters task
Sleep duration was negatively correlated with WM.
Vriend et al. (2012)
N = 32 Ages 8-12 years Community sample Canada
Correlational study
Actigraphy Sleep diary (parent) Parent report of sleep problems
Digit span modified from WISC-IV Spatial span modified from WRAML
No significant association between sleep duration or efficiency and WM (composite of auditory and spatial measures).
Vriend et al. (2013)
N = 32 Ages 8-12 Community sample Canada
Experimental sleep restriction and extension
Actigraphy Sleep diary (parent)
Digit span modified from WISC-IV & WRAML Finger Windows
Significant difference between WM performance after restricted compared to extended sleep. Differences with normal sleep not reported.
Wolfe et al. (2014)
N = 21 Ages 15-20 Videogame players Australia
Correlational study
Actigraphy Difference in operation span task performance between evening and morning
No relationship between sleep duration and WM performance next morning.
Abbreviations: AWMA = Automated Working Memory Assessment; BRIEF = Behaviour Rating Inventory of Executive Function; CCQT = Children’s ChronoType Questionnaire; fMRI = functional magnetic resonance imaging; MSLT = Multiple Sleep Latency Test; PSG = polysomnography; SES = socio-economic status; WISC-III/IV = Wechsler Intelligence Scale for Children, 3rd/4th edition; WM = working memory; WRAML = Wide Range Assessment of Memory and Learning Asterisks indicate studies which were identified in addition to the original PubMed/PsychInfo search.
31
Overview of associations The 15 selected studies were examined for their findings regarding relationships
between sleep and working memory variables. Only reported analyses are
represented in Table 2; it is likely that many more null results were found but not
reported (Ioannidis, Munafo, Fusar-Poli, Nosek, & David, 2014). In addition, all of
the studies using actigraphy to measure sleep duration also collected sleep diaries,
but only used the actigraphy data in their analyses as it was seen to be a more
objective measure.
Overall, eight of the 15 studies found no direct relationship between any measures
of sleep duration and working memory, while the other seven studies found a range
of relationships: positive and negative linear correlations, and inverted-U
relationships where working memory performance was reduced at both long and
short sleep durations.
32
Table 2: Results of selected studies
Objectively measured sleep duration
Reported sleep duration Experimental manipulation of sleep duration
Other sleep variable
Verbal WM
4 no association (Buckhalt, El-Sheikh, & Keller, 2007; Calhoun et al., 2012; Gruber et al., 2010; Wolfe et al., 2014) 1 negative correlation (Geiger, Achermann, & Jenni, 2010)
1 positive correlation (Cho et al., 2015) 2 negative correlations (Geiger et al., 2010; Van der Heijden, de Sonneville, & Swaab, 2013) 1 inverted U relationship (Gradisar, Terrill, Johnston, & Douglas, 2008)
3 no association (Beebe, Difrancesco, Tlustos, McNally, & Holland, 2009; Jiang et al., 2011; Sadeh, Gruber, & Raviv, 2003)
2 positive correlations with sleep efficiency (Buckhalt et al., 2007 - only at low SES; Steenari et al., 2003) 1 positive correlation with daytime sleepiness (Calhoun et al., 2012)
Visual WM
1 positive correlation (Steenari et al., 2003)
1 no association (Cho et al., 2015)
1 positive correlation with sleep efficiency (Steenari et al., 2003)
Combined verbal & visual WM
1 no association (Vriend et al., 2012)
1 no association (Konen, Dirk, & Schmiedek, 2015)
1 positive association (Vriend et al., 2013)
1 inverted U relationship with relative sleep duration (Konen et al., 2015)
Other 1 positive correlation with audiospatial WM (Steenari et al., 2003)
1 positive association with Digits Forward task (Sadeh et al., 2003) 1 association with brain activity (Beebe et al., 2009)
33
No relationship between sleep and working memory?
Over half of the selected studies (n = 8) reported no statistically significant
relationship between sleep and working memory (Beebe et al., 2009; Buckhalt et al.,
2007; Calhoun et al., 2012; Gruber et al., 2010; Jiang et al., 2011; Sadeh et al.,
2003; Vriend et al., 2012; Wolfe et al., 2014). While these findings would appear to
cast serious doubt on the hypothesis that sleep affects working memory in children,
there are some methodological issues which need to be considered.
Sample size has a direct effect on the probability of finding a statistically significant
result among the data (Field, 2013, p. 73). It would therefore be unlikely that Beebe
et al. (2009) would be able to detect any statistical relationship between sleep and
working memory, as their study included only six participants. The small sample
size was used because the main aim of the study was to examine brain activity
during working memory tasks, so the main outcome measures were functional
magnetic resonance imaging results. While no differences in WM performance were
found, consistent differences were noted in brain activation: the regions used to
complete WM tasks showed even greater activation after sleep deprivation, as if the
brain was working harder to achieve the same results. Likewise, Vriend et al. (2012)
reported a partial correlation (controlling for age) between sleep duration and
working memory scores of r = -0.28, which could be considered a medium effect
size (Cohen, 1992, p. 157) but which does not reach statistical significance because
there were only 32 participants.
In addition, reducing the variability within the sample will reduce the probability of
finding a significant correlation between measures (Goodwin & Leech, 2006). Five
of the selected studies excluded children with sleep, learning, behaviour and mental
34
health problems (Buckhalt et al., 2007; Gruber et al., 2010; Jiang et al., 2011; Sadeh
et al., 2003; Vriend et al., 2012) with only Vriend et al. (2012) acknowledging the
possibility that reduced variance may have affected their results.
There is a difference between long-term sleep habits and daily variation in the
duration of sleep each night. Wolfe et al. (2014) looked at a single night’s sleep
after being allowed to stay up late playing computer games, and their outcome
measure was the difference in working memory performance between the evening
and the morning tests. They found that sleep duration did not significantly mediate
this measure of working memory. They did, however, find that sleep duration
mediated the difference in sustained attention measures from evening to morning.
In these older participants (aged 15-20 years), it seems that short term variation in
sleep duration may influence attention, but no effect was found on working memory.
This study also suffers from a small sample size with low variance among
participants, so there may be smaller effects on WM that did not reach statistical
significance.
An additional problem with all of the studies finding no association between sleep
and working memory is that only linear relationships were tested. Two of the studies
which did find an association noted a skewed quadratic relationship: an inverted-U
shaped relationship in which both short and long sleep durations were linked to
poorer WM performance (Gradisar et al., 2008; Konen et al., 2015). It is possible
that some of these eight studies may have found an association between working
memory and sleep if non-linear relationships had been tested.
Finally, there may be important mediating factors which could mask any overall
relationship between sleep and working memory. For example, Buckhalt et al.
(2007) noted that socioeconomic status acted as a mediator: sleep efficiency was
35
correlated with WM performance but only in children from families of low
socioeconomic status. They hypothesised that high socioeconomic status brings
resilience against the effects of poor sleep. Weeknight or weekend measures may
also make a difference, particularly in older children when their weekend sleep
duration tends to be greater compared to weeknights (Gradisar et al., 2011).
Related to this finding, sleep duration is also greater during school holidays
compared to term times (Szymczak, Jasinska, Pawlak, & Zwierzykowska, 1993),
raising the possibility that the relationship between sleep and WM may vary
depending on school schedules. Interestingly, for the 10-year-olds in the Szymczak
et al. (1993) study, no significant seasonal variations in sleep duration were found,
suggesting that time of year should not affect studies of sleep in this age group.
A positive linear relationship between sleep and working memory?
Three studies showed a positive correlation: longer sleep durations were associated
with better working memory performance (Cho et al., 2015; Steenari et al., 2003;
Vriend et al., 2013). The samples in the first two of these studies showed the
opposite characteristics than those of the studies finding no relationship between
sleep and WM: these were relatively large sample sizes with variance maintained by
excluding very few children. Vriend et al. (2013) reported on the same sample as
their study from the previous year, but this time focused on the effects of
manipulating sleep duration rather than correlations between baseline data.
In fact, the sample size in Cho et al. (2015) was so large that they risked finding
statistically significant results for very small effects. To counteract this problem, Cho
et al. (2015) cited effect sizes, which adjust for these sample size biases (Field,
2013, p. 79). The authors confirmed that the effect sizes reported were Cohen’s d
36
(G. Roberts, personal communication, 28th April 2015). In this study, a positive
association between reported sleep duration on school nights and verbal working
memory was found (although with a wide confidence interval: d = 0.3; 95% CI 0.0 to
0.5, p = 0.02), but no such relationship with visual working memory (d = 0.0, 95% CI
-0.3 to 0.3, p = 1). One reason for this difference may be ceiling effects: the mean
performance on this standardised visual WM task was at the 75th centile, while the
mean performance of verbal WM was at the 63rd centile. This study is suggestive of
a relationship between sleep duration and working memory, but confidence in the
findings could be increased if the parental reports of sleep duration were
corroborated by objective measurement such as actigraphy.
Steenari et al. (2003) found that sleep duration, efficiency and latency were all
significantly correlated with visual and auditory working memory performance at the
higher levels of task difficulty. The tasks were rather unusual, using an auditory-
spatial n-back task rather than any measure of verbal WM. On closer inspection, it
is notable that the table of results presented gives correlation coefficients controlling
for age only. The authors found that visual and auditory WM performance at all
levels of difficulty was correlated with socioeconomic status. When they reanalysed
their data controlling for socioeconomic status, the effects of sleep mostly
disappeared and sleep duration was no longer associated with WM performance.
Unfortunately these general linear models are not reported in enough detail to know
exactly what analysis was done, why a repeated measures model was chosen when
all measures were only taken once, and how much variance in working memory
performance was explained by each factor (age, socioeconomic status, gender and
sleep variables).
Vriend et al. (2013) used an experimental design, taking a baseline and then asking
participants to restrict and extend their sleep by one hour for four weeknights in
37
each condition. The reporting of how successful participants were at changing their
sleep duration is inconsistent, stating that all participants achieved a minimum of 30
minutes’ sleep extension but not stating the range of sleep restriction achieved. A
key problem with the analysis in this study is that all comparisons are between the
long sleep and short sleep conditions. It would have been much more informative to
include the baseline condition as well, in order to identify any non-linear effects and
to analyse whether any changes are caused by sleep extension, sleep restriction, or
both. In addition, the authors give only combined data for both visual and verbal
WM tasks, making it impossible to tell whether the association with sleep duration is
significant for both modalities. The analysis provided is an ANOVA, in which WM
performance in the short-sleep condition is found to be significantly poorer than in
the long-sleep condition. No effect size is given, but from the means and standard
deviations provided, Cohen’s d can be calculated as 0.46 which would be
considered a medium effect size (Cohen, 1992, p. 157).
A negative linear relationship between sleep and working memory?
Both Geiger et al. (2010) and Van der Heijden et al. (2013) found that sleep duration
was negatively correlated with verbal working memory, i.e. children sleeping for
longer scored worse on verbal working memory tasks. Neither study included a
measure of visual working memory. Sample sizes were relatively large (60 and 333
participants respectively).
Geiger et al. (2010) report both actigraphic and parent-reported habitual sleep
timings, finding that both were associated with working memory scores.
Interestingly the correlations between sleep duration and WM reached significance
at the 0.05 level for weekends but not for weekdays, which perhaps reflects a more
38
natural sleep duration than weekdays which usually have more rigid bedtimes and
wake times. Geiger et al. (2010) report correlations adjusting for age and
socioeconomic status, with Pearson r values between 0.21 and 0.29 for the
correlations between WM and different measures of sleep duration, which is a
medium effect size (Cohen, 1992, p. 157). This effect size is particularly impressive
given the restricted sample variance in terms of socioeconomic status, and its skew
towards participants of higher socioeconomic status which might reduce the effects
of sleep on WM performance (Buckhalt et al., 2007).
Van der Heijden et al. (2013) were primarily interested in parental report of whether
their child tended to be more of an evening or a morning person, however they also
gathered a parentally-completed sleep diary for one week. While the response rate
was relatively high for the sleep diaries (80%), no analysis of non-responders is
provided so there may be bias within the sample. Working memory was assessed
using the Four Letters task from the Amsterdam Neuropsychological Tasks (De
Sonneville, 1999). Unfortunately this task confounds verbal and spatial WM, as the
participant has to judge whether the letter they just saw is both present and in the
right location within an array of four letters. Given the small number of possible
locations, the task has been classified as a verbal WM task for the purposes of this
review because the main load on recall is the letter itself. In this study, correlations
(adjusted for age, gender and socioeconomic status) between sleep duration and
WM were statistically significant at the 0.05 level but represented a relatively small
effect size (weekdays r = 0.18; weekends r = 0.16); this finding is explained by their
large sample size, which increases the significance of effect sizes.
In both these studies, a negative association between sleep and cognitive function
was found for a range of skills, not just working memory, suggesting that these
findings are not simply chance but represent either a real effect or systematic biases
39
within the studies.
An inverted U relationship between sleep and working memory?
Two studies (Gradisar et al., 2008; Konen et al., 2015) found an inverted U-shaped
relationship, in which children with medium sleep durations achieved higher working
memory scores than either long or short sleepers. Both found that the curve was
skewed towards shorter sleep, i.e. short sleep is associated with lower WM scores
than long sleep durations.
Gradisar et al. (2008) took a cross-sectional approach and divided their participants
into three groups for analysis: those getting less than eight hours’ sleep per
weeknight, those getting eight to nine hours and those getting over nine hours’ sleep
per weeknight. This method split the participants into roughly equal groups.
Analysis was carried out using ANCOVA, controlling for age, gender and vocabulary
score as a proxy for IQ. Socioeconomic status was not measured in this study. For
both measures of WM, there was a significant main effect of sleep duration group.
Pairwise comparisons showed that the differences between the medium and short
sleep groups were statistically significant (p < 0.05) and medium to large Cohen’s d
effect sizes were reported (d = 0.56 for Letter Number Sequencing and d = 0.92 for
Operational Span). Inspection of the means and standard deviations shows that the
long sleep group also had lower WM scores on both measures, but because the
pairwise comparisons did not reach statistical significance, the effect sizes were not
reported. They can be calculated, however (d = 0.37 for Letter Number Sequencing
and d = 0.17 for Operational Span), showing small to medium size effects according
to Cohen (1992). Correlation analyses looking for quadratic relationships were not
conducted, but from the data reported these might have fit the data better than a
40
linear model.
The other study finding a similar pattern was that of Konen et al. (2015), who took
the innovative approach of issuing their participants with smartphones, on which
participants were invited to answer questions on three occasions per day. The
times at which data were collected had been pre-agreed with the school, so that
teachers were aware when the children would be prompted to complete their data
collection and could plan these breaks into the day. Collecting data daily for a
month meant that the authors could identify each child’s usual sleep habits and any
nights when they departed from this pattern. While the sleep duration measures in
this study are self-report and not objective data, they may be more accurate than
most diary studies because the children were prompted every morning to report their
bedtime and waketime, rather than having a weekly diary which might be completed
retrospectively. The authors gave 2 WM tasks: one verbal and one spatial, but all of
the results report combined data rather than verbal and visual WM separately.
While Konen et al. (2015) found no overall between-subjects correlation between
working memory scores and reported sleep duration, they noted a relationship
between morning WM scores and the previous night’s sleep duration in comparison
to that child’s mean sleep duration. The findings showed that WM scores were
optimal on mornings when the child had an average duration of sleep the previous
night, and lower with longer or shorter sleep durations. No quantitative analysis of
the strength of this quadratic relationship was reported, but the change in r2 from
sleep duration was 3.1% including linear and quadratic relationships (T. Konen,
personal communication, 5 March 2015). This is a small effect size (Pearson’s r =
0.17). It is unlikely to be a clinically significant association, since even children
reporting that they slept six hours less than usual on the previous night showed a
working memory performance decrement of less than half a standard deviation (WM
41
accuracy at average sleep duration = 0.67, standard deviation = 0.28, WM accuracy
at six hours below average sleep duration estimated from graph = 0.56).
Weight of evidence assessment
When considering these studies together, it is important to take into account their
quality and relevance, so that some will be given greater weight than others. Gough
(2007) gives a framework for assigning weight of evidence (WoE) in systematic
literature reviews, which is used here. The criteria are set out in relation to this
specific review question, and then each study is judged against them (Table 3).
Weight of evidence A (Generic quality) – Does the study have a clear purpose
and a relevant method to achieve it? Is the study clearly described and are the
analyses appropriate?
Weight of evidence B (Relevance of method) – This review is centred on the
effects of sleep duration on working memory. Since this question involves a causal
link, experimental studies will be given greater weight than cross-sectional studies.
In addition, studies using actigraphy will be given greater weight as this captures
objective data in an ecologically valid way. Studies using polysomnography will
have lower weighting as they measure sleep in an artificial environment, and studies
using only sleep diaries or reported habits will have lower weighting as they
underestimate total sleep duration (particularly night-time wakings) (e.g. Holley, Hill,
& Stevenson, 2009). Studies using measures of verbal and visual working memory
and analysing these separately will be given greater weight than those using only
verbal WM measures, or reporting only combined visual and verbal WM scores.
42
Weight of evidence C (Relevance of focus) – Studies aimed specifically at
researching the relationship between sleep and working memory will be given
greater weight; those focused on other questions with this as a subsidiary focus will
be given lower weight. This review focuses particularly on children aged 9-10 years,
so studies including this age range will be given greater weight.
Weight of evidence D (Overall) – A judgement combining the previous three WoE
assessments to give an overall rating of the extent to which that study helps to
describe the effects of sleep duration on working memory in 9-10 year-old children.
43
Table 3: Weight of evidence analysis
Authors & year
WoE A
Generic quality
WoE B
Relevance of methodology
WoE C
Relevance of focus
WoE D
Overall
Beebe, Difrancesco, Tlustos, McNally, & Holland (2009) High Low Low Low
Buckhalt, El-Sheikh, & Keller (2007) High Medium-high Medium Medium
Calhoun et al. (2012) Medium Low Low Low
Cho et al. (2015) Medium Low-medium Medium-high Medium
Geiger, Achermann, & Jenni (2010) Medium-high Medium Medium Medium
Gradisar, Terrill, Johnston, & Douglas (2008) High Medium Medium Medium
Gruber et al. (2010) High Medium-high Medium-high High
Jiang et al. (2011) Low Medium Medium Low
Konen, Dirk, & Schmiedek (2015) High Low Medium Medium
Sadeh, Gruber, & Raviv (2003) High Medium-high High High
Steenari et al. (2003) Medium Medium-high High Medium
Van der Heijden, de Sonneville, & Swaab (2013) Medium-high Low Medium Medium
Vriend et al. (2012) Medium Medium Medium-high Medium
Vriend et al. (2013) Medium Medium-high Medium-high High
Wolfe et al. (2014) Medium Low-medium Low Low
44
All four of the studies with low WoE D ratings found no association between sleep
duration and memory. Of the three studies with high WoE D ratings, one (Vriend et
al., 2013) found a positive linear relationship between sleep duration and working
memory. The other two studies with high WoE D ratings found no significant
association between sleep duration and WM, but did report associations between
sleep and other related cognitive functions: short term memory (Sadeh et al., 2003)
and perceptual reasoning (Gruber et al., 2010). The remaining eight studies
received a medium WoE D rating, representing a mix of study designs and findings.
The WoE analysis does not systematically indicate a set of findings which are more
likely to answer the review question; studies of medium and high weightings provide
conflicting evidence about the existence and nature of any relationship between
sleep and working memory. This result may reflect the state of the evidence base in
this field, which draws from a wide range of methodological approaches and
participant groups. The inconclusive WoE analysis also suggests that relationships
between working memory and sleep duration in children may be less robust than in
adults, with more complexity in terms of mediating factors and mechanisms which
may change throughout child development. While this WoE analysis has not helped
to draw conclusions about how sleep affects WM in children, reviewing the evidence
in this way has highlighted some methodological aspects of research design which
could help guide future work towards answering the question.
Conclusions
The existing research looking specifically at the relationship between sleep and
working memory in children is small, although there are many related studies which
look at the relationships between sleep and other variables including other aspects
45
of cognition. The methodologies used are highly variable, including cross-sectional
studies, longitudinal studies and experimental studies manipulating sleep duration.
Subjective measures of sleep range from parent questionnaires to sleep diaries
(self- or parent-completed), and objective measures include actigraphy and
polysomnography. Measures of working memory also vary from study to study, with
many omitting measures of visual WM or combining visual and verbal results into a
single WM score. These variations in research design may contribute to the lack of
a clear relationship between sleep and working memory being found, although if
there was a robust association we might expect these different methodologies to
confirm it through triangulation of data.
Developmental issues complicate the issue further. Recent research has indicated
that the neural basis of working memory functions changes over the course of
childhood, with hippocampal areas subserving WM until the fronto-parietal network
matures. It is therefore entirely possible that the relationship between sleep and
WM may vary during development, if sleep affects brain areas differentially.
In conclusion, while many authors write as if there is a well established link between
sleep behaviour and working memory in children (e.g. Kopasz et al., 2010), this
literature review shows that the relationship is not clear at all. Results of published
studies range from finding no relationship between sleep and WM, to linear
relationships in both directions, to U-shaped relationships. Within the studies
involving an experimental manipulation, all showed effects of changing sleep
duration on at least one variable, but with no consistency about which aspects of
memory were impacted. There is therefore a clear role for future research in
clarifying whether sleep is reliably related to WM in children at different stages of
development, whether changes in sleep can produce changes in WM performance,
and the identification of factors which may mediate any effects of sleep on WM at
46
different ages.
Future research
This literature review leads to a number of recommendations for the design of future
research to help us understand the effects of sleep on working memory in children.
First, the sample to be studied needs to represent the range in the population.
Restricting variance in the sample by exclusion criteria such as learning difficulties,
behavioural difficulties or even sleep duration (Vriend et al., 2012 ; Vriend et al.,
2013 report two phases of the same study in which children who habitually slept less
than 8 hours or more than 12 hours per night were excluded) is likely to reduce the
effect size in a phenomenon which is already far from robust. In particular, limiting
the sample to children from families of higher socioeconomic status (SES) may
reduce the effect size if the findings of Buckhalt et al. (2007) can be generalised.
Proactive voluntary participation in research tends to skew the sample towards
higher SES participants (e.g. Damery et al., 2011), so one possible solution would
be to use a study design which could ethically use opt-out consent in order to
include the whole SES range.
Second, developmental factors need to be taken into account. Children of different
ages may well show different relationships between sleep and WM performance,
with one hypothesis being that the relationship might grow stronger as the child
matures and relies more upon fronto-parietal networks to undertake WM tasks.
Studies should therefore correct for age statistically or ensure that their sample(s)
are from narrow age ranges likely to be at a similar stage of brain development.
The measures used should include actigraphy for the most ecologically valid
47
objective way to measure sleep duration. In all the selected studies using both
objective and self- or parent-reports, the methods appear to measure different things
and the report measures overestimate sleep duration. Objective sleep duration has
repeatedly been found to bear no relation to subjective tiredness or daytime
sleepiness (Calhoun et al., 2012; Gruber et al., 2010), however individual
differences may explain this null finding as children experiencing sleep restriction
relative to their normal duration do report feeling more tired (Sadeh et al., 2003).
Future research should consider gathering and analysing both sleep diary and
actigraphy measures, in order to capture these different aspects of sleep.
Measures of working memory should include both visual and verbal WM, analysed
separately. Most studies have used measures of verbal WM only, or combined the
two into a single measure of WM, but when the two are analysed separately (as in
Cho et al., 2015) there may be differences between verbal and visual WM in their
relationship with sleep variables.
Finally, more longitudinal experimental studies would be helpful. As in the work of
Vriend et al. (2012; 2013), studies including an intervention to manipulate sleep
allow for a cross-sectional analysis of baseline data and also an analysis of
intervention effects. Studies of this nature can provide a better understanding of
causation than correlations alone.
48
Empirical study
49
Introduction
It is often stated that sleep has a well-established relationship with cognition, such that
shorter sleep duration is associated with poorer cognitive function (e.g. Beebe, 2012;
Durmer & Dinges, 2005; Ferrie et al., 2011; Killgore, 2010; Luber et al., 2008). This
paper aims to test this supposition, specifically examining sleep and working memory
(WM) in junior aged children.
A recent meta-analysis of studies involving total sleep deprivation in adults found a
robust effect for attention, processing speed, short term memory, working memory and
reasoning abilities (Lim & Dinges, 2010). Another meta-analysis, this time of studies
comparing adults with and without insomnia, showed significant effect sizes for some
cognitive skills (working memory, episodic memory, problem solving) but not others
(alertness, language and procedural memory) (Fortier-Brochu et al., 2012). An earlier
meta-analysis showed significant effects of sleep deprivation on cognitive functioning,
with even greater effects on mood (Pilcher & Huffcutt, 1996). These adult studies
suggest that there is a consistent correlation between reduced sleep duration and
reduced WM functioning.
In children, Astill, Van der Heijden, Van Ijzendoorn and Van Someren (2012)
conducted a meta-analysis of studies relating sleep duration and efficiency to cognitive
and behavioural variables. Unfortunately, the relatively small number of studies
available meant that this analysis collapsed data from a wide age range (5-12 years,
and including studies with participants whose age range overlapped with this range).
In addition, cognitive measures were collapsed into categories, e.g. “executive
functions,” within which Astill et al. (2012) included measures of skills that they
describe as inhibitory control, working memory and cognitive flexibility. The study
designs included both sleep deprivation and naturally occurring variations in sleep
50
duration, unlike the adult meta-analyses. Their findings were mixed but all effect sizes
were small (r = 0.02 to r = 0.1; r = 0.07 for executive functions). While the findings for
executive functions were statistically significant, the effect size is so small as to call into
question the authors’ conclusion that, “Executive functioning and the prefrontal
cortical circuits involved thus appear sensitive to sleep curtailment even early in
development and quite robustly so” (p. 1124). Focusing exclusively on school
performance, another meta-analysis also found a small effect size (r = 0.069) for sleep
duration (Dewald, Meijer, Oort, Kerkhof, & Bogels, 2010). The relationship between
sleep and cognition therefore seems to be less robust in children than in adults, and
these meta-analyses do not provide specific information about sleep and working
memory.
Working memory is a cognitive function of particular interest to educators. There is no
commonly accepted definition of WM, but Baddeley (2012) describes it as, “a
hypothetical limited capacity system that provides the temporary storage and
manipulation of information that is necessary for performing a wide range of
cognitive activities” (p. 7). Working memory is, therefore, distinct from short-term
memory which only requires information to be reproduced without manipulation, and
from long-term memory which holds information after it has been used for its original
purpose. Working memory has been found to predict future academic achievement
(Alloway & Alloway, 2010; Bull et al., 2008), especially in maths (Gathercole et al.,
2004), and has been implicated in the development of reading difficulties (Alloway,
2009; Gathercole, Alloway, Willis, & Adams, 2006).
Theoretically a link between sleep and working memory would be plausible, given the
shared neural substrates. During sleep in adults, prefrontal cortex is differentially
deactivated (Braun et al., 1997; Maquet, 2000; Muzur, Pace-Schott, & Hobson, 2002)
compared to other areas of the brain. More recently, research methods enabling an
51
analysis of functional connectivity have found that not only is the prefrontal cortex less
active during sleep, it is also decoupled from other areas of the brain (Horovitz et al.,
2009; Yoo, Gujar, Hu, Jolesz, & Walker, 2007).
This same brain area is particularly important in WM tasks, as part of a fronto-parietal
network (Rottschy et al., 2012). Animal and human studies show that lesions to the
dorsolateral prefrontal cortex result in working memory problems (Barbey, Koenigs, &
Grafman, 2013; Levy & Goldman-Rakic, 1999; Owen, Downes, Sahakian, Polkey, &
Robbins, 1990; Petrides, 2000). Non-invasive stimulation to the prefrontal cortex can
alter WM performance (Brunoni & Vanderhasselt, 2014). In addition we are starting to
gain some understanding of the neurochemical basis underlying WM functions, and
the genes contributing to individual differences in those systems (Blasi et al., 2015;
Gelao et al., 2014), lending further credence to the importance of networks involving
the prefrontal cortex in adult WM performance.
Studies using imaging methods to look at brain activity during WM tasks after sleep
deprivation in adults confirm that the fronto-parietal network reduces its activity in line
with reductions in WM performance (Chee et al., 2006; Mu et al., 2005). However, with
more demanding tasks, participants were able to use compensatory mechanisms and
showed higher activation in prefrontal cortex along with performance more similar to
before the sleep deprivation. (Chee & Choo, 2004; Drummond & Brown, 2001).
While the majority of the evidence would suggest that sleep duration is linked to
working memory performance in adults, through mechanisms involving the functioning
of the fronto-parietal network, the evidence for this link in children is much less clear.
Studies investigating relationships between sleep behaviour and working memory
performance in children have reported mixed results. Many have found no significant
52
association between the two, either through correlational analyses (Buckhalt et al.,
2007; Calhoun et al., 2012; Gruber et al., 2010; Konen et al., 2015; Vriend et al., 2012;
Wolfe et al., 2014) or by experimental manipulation of sleep duration (Beebe et al.,
2009; Jiang et al., 2011; Sadeh et al., 2003). Other studies have found an association
between sleep duration and WM performance, but without consistency in the nature of
the relationship. Some find a positive correlation (more sleep is related to better WM
scores; (Cho et al., 2015; Steenari et al., 2003; Vriend et al., 2013), others a negative
correlation (shorter sleep is related to better WM scores; (Geiger et al., 2010; Van der
Heijden et al., 2013) and still others an inverted-U relationship (optimal sleep is related
to better WM scores with worse performance at both longer and shorter sleep
durations; (Gradisar et al., 2008; Konen et al., 2015).
The neural development underlying the maturation of working memory skills may go
some way to explaining why the relationship between sleep and WM is less robust in
children. There is some evidence that the fronto-parietal network which subserves WM
performance in adults (Rottschy et al., 2012) takes a long time to reach maturity, and
that improvements in WM across childhood are related to the maturation of this
network (Darki & Klingberg, 2015; Tamnes et al., 2013). In the meantime, children
appear to make more use of hippocampal areas to perform WM tasks (Chaddock et
al., 2010; Finn, Sheridan, Kam, Hinshaw, & D'Esposito, 2010; Thomason et al., 2009).
It may therefore be that sleep affects working memory performance only as far as the
individual is using fronto-parietal networks to carry out the task. However, this
explanation cannot be the whole story, as the studies described above do not
represent a pattern of increasing correlation of WM with sleep duration as the
participants get older. In addition, there is evidence from studies of rodents (Ruskin et
al., 2004) and humans (Van Der Werf et al., 2009) that sleep duration affects
hippocampal functioning, so even if children were using hippocampus to undertake
WM tasks we might still expect sleep to affect functioning.
53
It is likely that methodological issues also contribute to the difference in effects seen in
children compared to adults. The adult studies included in the meta-analyses largely
involved sleep deprivation, either through insomnia or through experimental
manipulation, including total sleep deprivation for at least one night. It would be
ethically and practically difficult to use these approaches with children, so these studies
tend to involve very modest sleep deprivation or extension (up to one hour per night) or
naturally-occurring correlations between sleep and cognitive variables. Given that the
variation in sleep is smaller with these designs, it is unsurprising that the effect sizes on
cognition are smaller.
The current study was designed to further our understanding of the relationship
between sleep habits and working memory performance in children. The study design
allows for both correlational analysis and experimental effects. Rather than direct
manipulation of sleep duration (telling children to restrict or extend their sleep), a sleep
education programme was used in order to have a greater chance of long-lasting
effects. Children in a single year group were recruited, in order to minimise variation
due to age. Year 5 was chosen (age 9-10 years) in order to study children before the
onset of puberty (Laberge et al., 2001) but old enough to have an established sleeping
pattern without daytime naps (Thorleifsdottir et al., 2002) and to be taking more control
of their own behavioural habits (e.g. Such & Walker, 2004).
The key question for this research was, “what are the effects of sleep duration on
working memory in children aged 9-10 years?” Specific research questions included:
Is sleepiness, sleep duration or efficiency correlated with working memory
performance?
Does sleep education improve sleepiness, sleep duration or efficiency?
Does improved sleep correlate with improved working memory performance?
54
There were also methodological questions which could be addressed by this study,
including:
Does sleep diary information reflect actigraphy data in this age group?
Does wearing an actigraph systematically change the information reported in
sleep diaries?
55
Method
Study design
A quasi-experimental design was employed, in which half of the participating classes
acted as a control group and half undertook an intervention to improve sleep
behaviours. After the study had finished, the classes in the control group were
provided with the resources to undertake the intervention if they wished, but as there
was no expectation for them to do so they were considered a no-intervention rather
than a waitlist control group. This design enabled a cross-sectional analysis at
baseline, as well as a longitudinal analysis of experimental effects. The study received
ethical approval from University College London (see Appendix 1).
Data collection was designed around the length of the intervention. In the first week
(Time 0), baseline sleep diary data were collected. During the following week (Time 1)
the pre-intervention data were collected. Post-intervention data were collected five
weeks later (Time 2), leaving enough time between data points for the intervention
group to carry out the sleep education programme. Follow-up data were collected two
to three months after that (Time 3). Only term time weeks were counted, so some
classes had an extra week or two weeks between data collection points, if a school
holiday fell during that period. See Appendix 2 for the complete project schedule.
56
At Time 0, only sleep diary data were collected. This information was required in order
to give a baseline before the participants knew which of them would be wearing an
Actiwatch later in the study. At Times 1, 2, and 3 the full range of data were collected:
Sleep diaries
Actigraphy
Sleep knowledge
Working memory
Attention
Sleep diaries were given out on the Monday, to be filled in for seven days and
returned. Actiwatches were given out on the Monday, to be returned on the Friday.
The author attended the school during the relevant week at the school’s convenience,
and completed the group-administered measures with the whole class. Over the
course of that week, the author and a trained honorary assistant psychologist
completed the individually administered measures with each participant. The whole
data collection procedure usually took a day and a half for each class. At Time 3 an
additional questionnaire was given to participants in the intervention group, seeking
their feedback on the programme they had undertaken.
Demographic data on each participating child were collected from existing school
records:
Gender
Postcode, used to estimate socioeconomic status via the Income Deprivation
Affecting Children Index (IDACI, Department for Communities & Local
Government, 2011)
Most recent end-of-year National Curriculum levels in english and maths
Special educational needs status (on SEN register or not)
57
Measures
Sleep diaries
Week-long sleep diaries were produced on a single sheet of A4 paper (see Appendix
3). The diary sheets were adapted from those used by Tremaine et al. (2010),
including time into bed, time fell asleep, time woke up and sleepiness ratings.
Actigraphy
A set of 10 Actiwatch Spectrum devices and the relevant software with docking station
were loaned by the manufacturer (Philips Respironics, Bend, OR, USA) for the
duration of this study. Each actigraphy participant was given an information sheet (see
Appendix 4) to take home on each of the three occasions that they were asked to wear
the Actiwatch. As advised by the Philips representative, the children were instructed to
fasten the Actiwatch in the presence of the author onto whichever wrist felt more
comfortable, and to leave it there until collection at the end of the week. The epoch
length was set to one minute intervals, as this setting was the most commonly used in
other studies (Meltzer, Montgomery-Downs, Insana, & Walsh, 2012).
Ideally, participants would have worn the Actiwatch for a full seven days, as
recommended by Acebo et al. (1999). However, this was not possible in practice, as
the software would not operate with the laptop available and the devices therefore had
to be returned to the author’s home computer for uploading. Thus, data was collected
for four weeknights (Monday to Thursday). These may be the most valid data anyway,
since sleep start time and total time in bed differed at the weekend compared to
weeknights in Acebo et al.’s (1999) sample of school-aged children during term time.
58
Each Actiwatch was given out on the Monday and collected on the Friday of the data
collection week. The data was then uploaded and the device cleaned ready for the
next participant. Occasionally children left their Actiwatch at home, so the parent was
asked to bring it into school or, if that was not possible, the author went to the house to
collect the device.
The data were viewed using Philips Actiware software version 6.0.0. Sleep onset was
defined as 10 minutes of immobility, which is the standard algorithm in this software.
To reduce the probability of wake after sleep onset being over-identified (Meltzer et al.,
2012) the wake threshold selection was set to “high”.
Sleep knowledge
Children’s knowledge about sleep was measured using a 10-item questionnaire (see
Appendix 5) based on Benveniste, Thompson & Blunden’s (2012) tool, which was
developed for children aged 9-15 years. Adaptations were made to the original
questionnaire. First, item 10 (“how many hours should you sleep every night to be
healthy?”) was turned from an open question into a closed one, in line with the other
items in the questionnaire. This necessitated a further change so that there were not
two questions about sleep length: item 6 (“I need more than 9 hours of sleep each
night”) was changed to, “children need less sleep than their parents.” Finally, item 8
(“boys have deeper sleep than girls”) was moved up to become item 2, in order to
avoid the existing sequence of five questions with, “true,” as the correct answer.
All of the items in the questionnaire were cross-referenced with the teaching materials
provided for the intervention, to ensure that the content needed to answer the
questions correctly was included in the lessons. The questionnaire was administered
to whole classes at once, and the researcher read each question aloud.
59
Working memory
Originally the study was designed to capture all cognitive data with group-administered
measures, in order to reduce disruption to the classes’ schedules and to reduce the
total time required for data collection. The Automated Working Memory Assessment
(AWMA) 2 was selected, as this web-based version could be completed by groups of
children at the same time in the school IT suite or on laptops in the classroom.
However, this product was withdrawn by the publisher in September 2013, just before
the data collection for this study began. The previous version of the AWMA (Alloway,
2007) therefore had to be substituted, and the publisher would only supply one CD-
ROM, necessitating individual testing on a laptop that had the software loaded.
This technical difficulty meant that the working memory tests had to be very selective
as there would not be time to undertake extensive testing with each child: the complete
AWMA includes three tests each of visual short term memory, verbal short term
memory, visual working memory and verbal working memory. One test of visual
working memory (Mr X) and one test of auditory working memory (Backward Digit
Recall) were chosen, to prioritise tests of working memory rather than short term
memory and to cover both visual and verbal modalities since these are theoretically
and empirically separable (e.g. Baddeley, 2012). These particular subtests were
chosen as the Pearson representative advised that another research group was using
them when they required a shorter administration time (this research is now published:
Dunning & Holmes, 2014). The switch from group to individual administration also
necessitated additional researcher time, and it was extremely fortunate that an
honorary assistant psychologist was available and could commit to one day per week
of data collection for the duration of the project.
60
Attention
To enable analysis of whether any effects of sleep on working memory were mediated
by attention (as suggested by Steenari et al., 2003), group measures of auditory and
visual attention were administered to the whole class at once. Auditory attention was
measured using the Score! subtest from the Test of Everyday Attention for Children
(TEA-Ch: Manly, Robertson, Anderson, & Nimmo-Smith, 1998). Although the reliability
is not quite as high as for the other test of sustained auditory attention in this battery
(Code Transmission), it was the only tool that could be adapted for group
administration as the responses can be written and do not have to be scored in real
time. The Score! test requires participants to listen to a series of tones at varied
intervals and keep track of how many were heard. After using this test with the pilot
class, instructions were explicitly given each time that the children should keep count in
their heads, and that no speaking aloud, tally marks or counting on fingers were
allowed. Observation confirmed that children did stick to these rules.
Visual attention was assessed using the d2 Test of Attention (Brickenkamp & Zilmer,
1998), which was developed in Germany. It is a cancellation test, requiring participants
to mark all the items fulfilling the criteria (letter d with two marks either above or below
it), and none of the distractor items. One child with dyslexia requested to complete his
d2 tests individually as he found it difficult, and this was arranged. The Concentration
Performance measure from the d2 task was used (rather than Total Number – Errors)
in order to ensure that credit was given only for the number of items endorsed and not
for skipping over large numbers of items.
61
Intervention
A universal rather than targeted intervention was used for this study, on the
assumption that the prevalence of sleep problems is high so most children would
benefit from improving their sleep habits (e.g. Spruyt, O'Brien, Cluydts, Verleye, &
Ferri, 2005). A universal programme also had the ethical advantage that all children in
a class could participate, without individuals being excluded because their parents had
not returned consent forms or because they did not meet criteria on screening for poor
sleep.
There are few universal sleep interventions available for primary aged children. Sleep
Well Be Well has been trialled in Australia (Quach et al., 2013) but it is targeted at
those who report sleep difficulties on a screening questionnaire, and is an
individualised programme requiring relatively high levels of resourcing (three 1-1
contacts per family). Moseley & Gradisar (2009) used a group-based universal
intervention in schools, but their programme involved older children (15-year-olds).
Reut Gruber’s group at McGill University have been working for some years on
developing their Sleep for Success programme, however only a video has been made
publicly available so far; the materials have not been published and there are no
articles describing or evaluating the intervention in peer-reviewed journals.
The most relevant intervention for this study was the Australian Centre for Education in
Sleep (ACES) programme (Blunden, Kira, Hull, & Maddison, 2013; Blunden, 2007a;
Blunden, 2007b; Kira, Maddison, Hull, Blunden, & Olds, 2013). This programme
comprises a set of sleep education sessions to be delivered to whole classes. While
there is more evaluation data on the adolescent programme, a junior programme for
younger children has also been developed. The lead author, Sarah Blunden, kindly
allowed these ACES Junior materials to be used in this study free of charge. The
62
materials were adapted for UK use and these adaptations were approved by Dr
Blunden (see Appendix 6).
The final intervention consisted of three lessons, a project and a parent booklet. The
Theory of Planned Behaviour (Ajzen & Fishbein, 1980) was the theoretical
underpinning for promoting behaviour change, as children were asked to consider their
current sleeping habits and write down any changes that they intended to make.
Lesson 1 covered normal sleep, lesson 2 was about sleep hygiene and lesson 3
looked at sleep problems. The project was less prescriptive but ideas were provided in
the teachers’ manual and the aim was to consolidate children’s learning from the
lessons. The parent booklet was provided because the sleep patterns of children 9-10
years old are still very much influenced by their parents (Meltzer & Montgomery-
Downs, 2011) and would be unlikely to change without parental support.
A training session for the teachers was held in November 2013. They were taken
through the content of the PowerPoint presentations, teachers’ manual, student
workbook and parent booklet, and discussed together any queries and the details of
how they would deliver the intervention in their classes. These materials were then
given out to the teachers who were to be in the intervention group, while the teachers
in the control group did not receive the materials until after all their data had been
collected.
All four teachers in the intervention group reported that they delivered the three lessons
and sent the parent booklets home. Some teachers reported omitting some of the
practical activities in the lesson plans, but all presented the PowerPoint slides and
used the workbooks. Although no observations of the lessons were carried out, lesson
delivery was checked by inspecting the children’s workbooks, which were completed in
all four intervention classes. Two classes did not complete a project, one invited
63
children to complete their project for homework (10 children did so) and one class
completed the project in a further classroom session. Treatment fidelity was therefore
an issue in this study.
Participants
Recruitment
The key outcome measure used to calculate the number of participants required was
extension in sleep time, as this outcome would show whether the intervention had
worked to increase the amount of sleep time children were getting. It was not possible
to use existing reports of sleep education programmes, as they either had older
participants or were not successful in achieving a change in sleep duration. Instead
Sadeh et al.’s (2003) study was used as a benchmark, as the children in this study
were approximately the same age as the current sample, and they were successful in
achieving an extension of sleep time over three successive nights in 62% of
participants, as measured by actigraphy. Using G*Power 3.1.3 software, and a test-
retest reliability of 0.6 (Sadeh, Dahl, Shahar, & Rosenblat-Stein, 2009), the calculated
repeated-measures effect size (dz) for the Sadeh et al. (2003) study was 0.78
(baseline mean = 487.7 minutes, s.d = 43 minutes; extension mean = 516.9 minutes,
s.d. = 40.2 minutes). To detect an effect of this size with 0.8 power in this study would
require 15 participants in the intervention group (the effect in this case would be pre-
post intervention change in sleep duration, rather than comparison to the control
group).
This power calculation was cross-referenced with correlational studies. Steenari et al.
(2003) found a correlation between sleep duration and the hardest auditory WM task of
r = 0.298 and for the visual WM task r = -0.325. G*Power 3.1.3 software calculated
64
that a total of 83 and 69 participants respectively would be required to detect effects of
these magnitudes with 0.8 power. Similarly, Vriend et al. (2012) found a correlation
between sleep duration and WM of r = -0.28, an effect size which would require 95
participants using the same parameters.
Given that some dropout from the study is to be expected (e.g. Quach, Hiscock,
Ukoumunne, & Wake, 2011, had partial or no data from 26% of the families in their
sleep education trial), at least three classes were to be recruited to each group. This
would enable actigraphy data to be collected from 30 children in the intervention group
(10 children in each class could have an Actiwatch) and give a good chance of
detecting an extension in sleep duration of similar magnitude to Sadeh et al. A study
of this size would likely involve a total of 150-180 children depending on class sizes
(six classes of 25-30 pupils), thus providing enough power to detect correlational effect
sizes even with some dropout.
Initial recruitment was carried out in September 2013 via the planning meetings that
each school in the author’s employing Local Authority holds annually with their link
educational psychologist (EP). All the EPs took along a flyer (see Appendix 7) and
gave it to each primary or junior school’s Special Educational Needs Co-ordinator (n =
50), asking for it to be passed on to the Year 5 teacher(s).
Nine schools contacted the author to express interest in becoming part of the study.
Two dropped out soon after receiving the teacher information sheet, in September and
November of 2013. This left seven schools who returned a teacher consent form (see
Appendix 8) signed by the head teacher, of which one had two parallel Year 5
teachers. After the project schedule had been arranged and the Time 0 sleep diaries
completed by the first class in that school, the second class joined the study, making
eight participating classes. All eight classes continued to be part of the study until the
65
end of data collection. Although the study design required only six classes, all eight
were accepted into the study for ethical reasons and to increase the chances of
achieving the numbers of participants indicated by the power analyses.
The eight participating teachers were given information sheets to hand out to each
member of their class (see Appendix 9). This sheet explained the study and the child’s
right to opt out of it. One child in the intervention group opted out of the data collection,
and in the control group a teacher decided not to invite one child to join the study as he
would not have been able to complete the measures. Both of these non-participants
had a diagnosis of autistic spectrum disorder. All the other children remained in the
study, although at each time point a small number were not in school on the day of
data collection.
66
Figure 5: Flow diagram of participation throughout the study
Schools approached
50 schools
Did not respond
41 schools
Tim
e 1
Schools requested further information
9 schools
Total in study 107 children
Sleep diaries n = 19
Actigraphy n = 37
Working memory n = 104
Knowledge/attention n = 102
Declined to participate
2 schools Agreed to participate
7 schools
7 classes
Tim
e 2
Allocated to control group
3 schools
3 classes
84 children
Total in study 113 children
Sleep diaries n = 55
Actigraphy n = 29
Working memory n = 110
Knowledge/attention n = 111
Allocated to intervention group
4 schools
4 classes
108 children
Declined to participate
1 child Excluded from study by teacher
1 child
Total in study 107 children
Sleep diaries n = 20
Actigraphy n = 25
Working memory n = 104
Knowledge/attention n = 100
Total in study 84 children
Sleep diaries n = 63
Rec
ruit
men
t
All
oca
tio
n
Tim
e 3
T
ime
0
Total in study 107 children
Sleep diaries n = 55
Total in study 113 children
Sleep diaries n = 76
Actigraphy n = 37
Working memory n = 111
Knowledge/attention n = 107
Total in study 113 children
Sleep diaries n = 55
Actigraphy n = 31
Working memory n = 105
Knowledge/attention n = 107
Additional class joined study
3 schools
4 classes
113 children
Total in study 107 children
Sleep diaries n = 33
Actigraphy n = 23
Working memory n = 102
Knowledge/attention n = 99
67
Assignment to control or intervention group
The participating teachers were contacted in October 2013 to ask if any of them would
be prepared to act as a pilot class, and one volunteered. This class began their data
collection later that month and formed the first class in the control group. At the
training session, the participating teachers were shown the schedule and asked to
decide among themselves which classes wished to begin on which dates. After this
timetable had been agreed, the first three classes were told that they would join the
control group, and would have time to deliver the intervention in the summer term after
their data collection was complete. The later four classes were told that they would
form the intervention group. This approach was taken in order that all classes in the
study could deliver the intervention within the academic year. If any of the control
classes had started on later dates, they would have finished the data collection phase
too late to offer the intervention to the children who had participated.
Characteristics of control and intervention groups
Demographic information about each participating child was provided by the school.
An estimate of socio-economic status was made from each child’s postcode: the
Income Deprivation Affecting Children Index (IDACI) score is the proportion of children
aged 0-15 in that area who lived in income-deprived households at the last census in
2010 (Department for Communities & Local Government, 2011). Therefore the higher
the IDACI score, the more deprived the neighbourhood where the child lives. These
scores were obtained by entering each child’s postcode into the UK government’s
quick checker: http://www.education.gov.uk/cgi-bin/inyourarea/idaci.pl For
comparison to national data, deciles have been calculated from the rank ordering
given to each IDACI score (Table 4).
68
Table 4: IDACI scores with corresponding ranks and deciles
IDACI score range Rank order range Decile
0 - 0.0344 29232 – 32482 1 (least deprived)
0.0345 – 0.0542 25984 – 29233 2
0.0543 – 0.0766 22736 – 25983 3
0.0767 - 0.1048 19488 – 22735 4
0.1049 – 0.1429 16240 – 19487 5
0.1430 – 0.1921 12992 – 16239 6
0.1922 – 0.2563 9744 – 12991 7
0.2564 – 0.3413 6496 – 9743 8
0.3414 – 0.4564 3248 – 6495 9
0.4565 – 1 1 – 3247 10 (most deprived)
Children’s SEN status was also provided by the school. The percentage of children
with SEN includes children on the SEN register at any level (School Action, Action Plus
or Statement). Attainment scores are based on teacher assessment at the end of the
previous school year (summer 2013). The points equate to National Curriculum levels,
and in England children are expected to achieve 21 points (also known as Level 3b) at
this stage in their education (end of Year 4, aged 9).
Table 5: Participant demographics
Number of
children
% boys Mean IDACI score
% SEN Mean reading score
Mean writing score
Mean maths score
Control 113 57.52 0.168 20.35 23 21 23
Intervention 107 49.53 0.372 16.82 21 20 21
Total 220 53.64 0.269 18.64 22 21 22
Overall the participants had relatively low socioeconomic status compared to other
children in the UK (mean IDACI score within the 8th most deprived decile), which
reflects the geographical area from which they were recruited. Despite this
disadvantage, the children in this study were achieving at nationally expected levels.
69
The intervention group lived in more deprived areas, had a lower proportion of boys,
were working at a slightly lower academic level than the control group, and had a lower
proportion of children identified as having special educational needs.
The control group included two schools (three classes in this study) which were rated
outstanding at their last Ofsted inspection, along with one requiring improvement. The
intervention group included two schools rated good, one requiring improvement and
one inadequate.
Participation in sleep diary data collection
Table 6: Numbers of sleep diaries completed
Class Number of children
participating
Sleep diaries
returned Time 0
Sleep diaries
returned Time 1
Sleep diaries
returned Time 2
Sleep diaries
returned Time 3
Control 1 31 24 27 14 16
Control 2 23 17 14 11 12
Control 3 30 22 21 16 14
Control 4 29 - 14 14 13
Total control 113 63 (75%) 76 (67.2%) 55 (48.7%) 55 (48.7%)
Intervention 1 25 15 0 4 7
Intervention 2 28 21 18 14 25
Intervention 3 27 6 0 0 0
Intervention 4 27 13 1 2 1
Total
intervention 107 55 (51.4%) 19 (17.8%) 20 (18.7%) 33 (30.8%)
Overall total 220 118 (61.8%) 95 (43.2%) 75 (34.1%) 88 (40.0%)
As Table 6 shows, participation in completing the sleep diaries was highly variable
across classes and declined over time. In particular, three of the four intervention
classes returned little or no diary data after the first week of the study. For individuals,
70
diary returns were not consistent, for example a child might have completed a diary at
Times 1 and 3 but not at Time 2. Complete diary data sets (returns at all four time
points) were available for only seven children in the intervention group and 20 children
in the control group.
Participation in Actiwatch data collection
The information sheet given to all children in participating classes included an opt-in
consent form for children wishing to volunteer to collect Actiwatch data. Where more
than 10 consent forms were received, a random selection was made. The participation
rates were highly variable, as shown in Table 7.
Table 7: Numbers of Actiwatches used
Class Consent forms received
Actiwatches taken Time 1
Actiwatches taken Time 2
Actiwatches taken Time 3
Control 1 28 10 10 10
Control 2 15 10 9 (1 withdrew) 8 (1 withdrew, 1
off sick)
Control 3 13 10 10 10
Control 4 8 8 - 7 (1 off sick)
Total control 64 38 29 35
Intervention 1 15 10 9 (1 watch lost) 10
Intervention 2 27 10 9 (1 watch lost) 9 (1 off sick)
Intervention 3 12 9 (1 watch lost) 5 (4 withdrew) 5 (4 withdrew)
Intervention 4 10 9 (1 watch lost) 6 (3 withdrew) 4 (5 withdrew)
Total intervention 64 38 29 28
Control class 4 were unable to collect Actiwatch data at Time 2 because they joined in
the schedule with control class 3 (in the same school). It was possible to find a spare
week in the schedule at the beginning and end of the programme, but at Time 2 their
data had to be collected on the same week as class 3, and there was only one set of
71
Actiwatches available.
One child in intervention class 2 lost their Actiwatch at Time 2. It took several weeks to
replace, and therefore intervention classes 3 and 4 started with only nine Actiwatches.
In three classes, some children chose not to continue wearing the Actiwatch after the
first week. They were not asked to give reasons for withdrawing their consent, but
several children spontaneously commented that they had found the Actiwatch
uncomfortable. One said that she had simply forgotten to wear it.
In addition, sometimes Actiwatches were returned with unusable data for some or all of
the nights. This problem may have been caused in some cases by children leaving the
watch off-wrist, and in other cases by technical faults with the devices; it is not possible
to identify which was the reason in each case. Mean sleep durations were calculated,
which in some cases meant a single night of usable data.
Table 8: Numbers of children with usable Actiwatch data
Class Number of children with usable Actiwatch data at Time 1
Number of children with usable Actiwatch data at Time 2
Number of children with usable Actiwatch data at Time 3
Control 1 9 10 10
Control 2 10 9 8
Control 3 10 10 8
Control 4 8 - 5
Total control 37 29 31
Intervention 1 9 7 8
Intervention 2 10 8 6
Intervention 3 9 4 5
Intervention 4 9 6 4
Total intervention 37 25 23
72
Results
Statistical analysis
Data analysis was undertaken using IBM SPSS Version 22 software.
Where data were not normally distributed, non-parametric tests were used to compare
means. When correlations and partial correlations needed to be calculated for non-
normal data, bootstrapping was used to calculate robust 95% confidence intervals
(CIs) (e.g. Field, 2013). Bootstrapping was used in preference to non-parametric
correlations in order to allow direct comparisons with partial correlations, for which
there is no non-parametric alternative in SPSS.
In bootstrapping, the actual data are used as a population from which random samples
are drawn and analysed (in this study, the samples were analysed for Pearson
correlations, denoted by r). The random samples are all the same size as the original,
but with replacement (i.e. the same value may appear more than once while another
value is excluded from the sample). This process allows an estimation of the sample
distribution for non-normally distributed data. The 95% CI tells us the range within
which 95% of the results from the random samples of the whole dataset fell. In this
study bias-corrected and accelerated bootstrapping was used, with 1000 samples.
As SPSS does not bootstrap the p-values, these are not reported and instead
significance is interpreted directly from the correlation CIs. If the 95% CI includes zero,
the probability of the null hypothesis being true is greater than 0.05, and the result is
therefore taken to be non-significant (Wood, 2005). This approach leads to the same
conclusions as conventional significance testing (Smith & Morris, 2015), making it an
equivalent way to analyse non-normally distributed data.
73
Baseline measures
Sleep duration
The sleep diaries yielded three measures of time asleep. Time in bed is the whole
period between the recorded time of getting into bed and the recorded time of waking
the next morning. Sleep duration is the whole period between the recorded time of
falling asleep and the recorded time of waking the next morning. Total time asleep is
sleep duration minus the amount of time the child said they were awake during the
night (wake after sleep onset).
Actigraphy yielded two measures of time asleep: time in bed and total time asleep,
which is time in bed minus time spent awake during the night.
All measures of sleep duration in this study are given as a mean, calculated over the
available weeknights (Monday to Thursday) for each time point. Sundays were not
included because several weeks within the study included a Bank Holiday Monday, so
some Sundays were not a schoolnight.
74
Table 9: Sleep duration at baseline
Mean duration (first data collection point)
Diary (time in
bed)
Diary (sleep
duration)
Diary (total time asleep)
Actigraphy (time in bed)
Actigraphy (total time asleep)
N 132 132 132 74 74
Mean 10 hrs 28 mins
9 hrs 41 min
9 hrs 37 mins
9 hrs 26 mins 8 hrs 9 mins
SD (minutes) 42.96 mins
47.17 mins
49.60 mins
33.07 mins 37.40 mins
Less than 7 hours
0 2 (1.5%) 2 (1.5%) 0 3 (4.1%)
7 hrs – 7 hrs 59 mins
0 2 (1.5%) 2 (1.5%) 0 22 (29.7%)
8 hrs – 8 hrs 59 mins
3 (2.3%)
15 (11.4%)
18 (13.6%)
18 (24.3%)
44 (59.5%)
9 hrs – 9 hrs 59 mins
28 (21.2%)
67 (50.8%)
70 (53.0%)
47 (63.5%)
5 (6.8%)
10 hrs – 10 hrs 59 mins
66 (50.0%)
43 (32.6%)
38 (28.8%)
8 (10.8%)
0
11 hrs – 11 hrs 59 mins
33 (25.0%)
3 (2.3%)
2 (1.5%)
1 (1.4%)
0
More than 12 hours
2 (1.5%) 0 0 0 0
The usual recommended amount of sleep for this age group is 9-11 hours (Matricciani,
Olds, Blunden, Rigney, & Williams, 2012). In their meta-analysis, sleep duration was
operationalised as, “the time between falling asleep and waking,” (p. 554). In the
current sample, most children were therefore getting enough sleep before the
intervention according to their diary data (Table 9). Looking at the actigraphy data,
approximately a quarter of children were averaging less than nine hours in bed per
weeknight. If night-time wakings are subtracted, giving total time asleep rather than
time in bed, less than 7% of children were getting more than nine hours of sleep –
75
however, actigraphic time asleep is not the metric against which most
recommendations are made. The mean total time asleep as measured by actigraphy
is almost exactly the same as the baselines reported by Sadeh et al. (2003): their
samples had a mean of 8 hours 8 minutes (sleep extension group) and 8 hours 9
minutes (sleep restriction group).
Using independent samples t-tests, no difference was found between the sleep
duration of boys and girls on any of these measures at the first data collection point
(diary time in bed, t = -1.071, df = 130, p = 0.286; diary sleep duration t = -0.275, df =
130, p = 0.784; diary total time asleep, t = -0.017, df = 130, p = 0.986; actigraphy time
in bed, t = -0.318, df = 72, p = 0.752; actigraphy total time asleep, t = -0.373, df = 72, p
= 0.463).
Pearson’s correlation coefficients between these measures of sleep duration and
socioeconomic status were calculated, with higher deprivation being significantly
associated with shorter sleep for diary sleep duration (r = -0.25, p = 0.005) and
actigraphy time in bed (r = -0.28, p = 0.016), but not for diary time in bed (r = -0.14, p =
0.113) or actigraphy time asleep (r = -0.15, p = 0.197).
As diary-reported sleep duration and total time asleep were so similar, because
children reported hardly any wake after sleep onset, all further analyses use total time
asleep only. This measure is analogous to the actigraphic measure of total time
asleep.
Sleep efficiency
Mean sleep efficiency (percentage of total time in bed spent asleep) measured by
76
actigraphy was calculated across the four weeknights at the first data collection point (n
= 74, mean = 86.58%, SD = 4.77%). This finding is consistent with other studies citing
sleep efficiency data for this age range (e.g. Steenari et al., 2003, 86.5%; Vriend et al.,
2012, 85.5%).The average amount of wake after sleep onset, as measured by
actigraphy, was almost an hour, with wide variation between children (n = 74, mean =
57 minutes, SD = 27 minutes). There was no relationship between sleep efficiency
and time in bed (r = -0.07, p = 0.540), but a strong relationship between sleep
efficiency and time asleep (r = 0.65, p <0.001). These findings are consistent with the
correlations between the same actigraphy measures reported by Holley et al. (2009)
for 6-11 year olds in the UK.
Working memory
Performance on the two working memory tasks at the first data collection point was
similar to that of the standardisation sample (mean = 100, SD = 15).
Table 10: Baseline working memory scores
Mean raw score
SD raw score Mean standard score
SD standard score
Visual WM 12.7 4.765 97.58 14.3
Verbal WM 11.6 4.093 101.2 13.76
As the working memory and socioeconomic status measures were not normally
distributed according to Kolmogorov-Smirnoff tests (verbal WM D(215) = 0.126, p
<0.001, visual WM D(215) = 0.091, p <0.001; IDACI D(215) = 0.123, p <0.001),
bootstrapping was used. The correlation between verbal and visual working memory
measures at the first time point was surprisingly low (n = 215, r = 0.15, CI -0.018 to
0.307) compared to another study using exactly the same tests of working memory
(Alloway, Gathercole, & Pickering, 2006, r = 0.35). Socioeconomic status was
77
correlated with working memory, but in different directions. For visual working memory
higher SES was related to higher scores (n = 215, r = -0.13, CI –0.05 to -0.244), but for
verbal working memory it was children of lower SES who did best (n = 215, r = 0.236,
CI 0.096 to 0.393). The effect sizes are rather small, consistent with other studies
showing that socioeconomic status tends not to be strongly related to measures of
working memory in children (e.g. Alloway & Alloway, 2010; Engel, Santos, &
Gathercole, 2008).
Working memory and sleep
At Time 1, 95 children had returned sleep diaries and also completed tests of working
memory. One outlier case (a child reporting less than five hours’ total time asleep) was
removed from this part of the analysis. Even with the outlier removed, most of the
measures reached significance on the Kolmogorov-Smirnoff tests of normality (Table
11) so bootstrapping was used.
Table 11: Tests of normality for working memory and diary data at Time 1
D d.f. p
Diary time in bed 0.073 94 >0.2
Diary time asleep 0.103 94 0.016
Diary sleepiness 0.116 94 0.003
Verbal WM 0.142 94 <0.001
Visual WM 0.123 94 0.001
As shown in Table 12, Pearson’s correlation coefficients showed that verbal working
memory was negatively correlated with reported time in bed and total time asleep, but
not with sleepiness ratings. Visual working memory was not significantly correlated
with any diary measure. It is notable that, in this subset of children who had submitted
sleep diaries, there was no correlation between their verbal and visual WM scores.
78
When socioeconomic status was entered as a control variable, all partial correlations
between sleep diary and WM measures were non-significant.
Table 12: Correlations between sleep diary and working memory variables at Time 1
Time in bed Total time
asleep Sleepiness Verbal WM Visual WM
Time in bed 1 r = 0.830 CI: 0.749 to 0.888
r = -0.035 CI: -0.220 to 0.154
r = -0.223 CI: -0.375 to -0.073
r = 0.057 CI: -0.133 to 0.243
Total time asleep
r = 0.820 CI: 0.740 to 0.876
1 r = 0.070 CI: -0.122 to 0.264
r = -0.224 CI: -0.375 to -0.073
r = 0.097 CI: -0.076 to 0.266
Sleepiness r = -0.004 CI: -0.193 to 0.193
r = 0.095 CI: -0.113 to 0.290
1 r = -0.074 CI: -0.231 to 0.105
r = -0.021 CI: -0.208 to 0.165
Verbal WM r = -0.154 CI: -0.333 to 0.014
r = -0.174 CI: -0.005 to 0.085
r = -0.104 CI = -0.255 to 0.032
1 r = -0.026 CI: -0.254 to 0.208
Visual WM r = 0.028 CI: -0.148 to 0.194
r = 0.077 CI: -0.101 to 0.252
r = -0.011 CI: -0.204 to 0.198
r <0.001 CI: -0.237 to 0.271
1
Data in the top right of the table are bivariate correlations; data in the bottom left are
partial correlations controlling for socioeconomic status.
Also at Time 1, both actigraphy and working memory data were available for 74
children (Table 13). One child’s data showed that they averaged over 11 hours in bed,
which is an outlier and was removed from this part of the analysis. This exclusion
normalised the data for the actigraphy measures, although the scores for working
memory remained non-normally distributed (verbal D(73) = 0.13, p = 0.004; visual
D(73) = 0.125, p = 0.006) so bootstrapping was used. Pearson’s correlation
coefficients showed that none of the actigraphy measures were significantly correlated
with either working memory measure, although with SES controlled for, a small but
significant correlation was found between sleep efficiency and visual working memory.
79
Table 13: Correlations between sleep actigraphy and WM variables at Time 1
Time in bed Time asleep Sleep efficiency
Verbal WM Visual WM
Time in bed 1 r = 0.672 CI: 0.524 to 0.796
r = -0.040 CI: -0.299 to 0.241
r = -0.201 CI: -0.403 to 0.021
r = -0.046 CI: -0.240 to 0.155
Time asleep
r = 0.667 CI: 0.522 to 0.787
1 r = 0.702 CI: 0.579 to 0.801
r = 0.015 CI: -0.219 to 0.256
r = 0.100 CI: -0.068 to 0.260
Sleep efficiency
r = -0.024 CI: -0.258 to 0.246
r = 0.718 CI: 0.593 to 0.814
1 r = 0.206 CI: -0.011 to 0.410
r = 0.185 CI: -0.007 to 0.373
Verbal WM r = -0.152 CI: -0.356 to 0.059
r = 0.045 CI: -0.205 to 0.278
r = 0.198 CI: -0.057 to 0.437
1 r = 0.103 CI: -0.146 to 0.350
Visual WM r = -0.129 CI: -0.325 to 0.069
r = 0.067 CI: -0.104 to 0.255
r = 0.212 CI: 0.021 to 0.380
r = 0.176 CI: -0.089 to 0.408
1
Data in the top right of the table are bivariate correlations; data in the bottom left are
partial correlations controlling for socioeconomic status.
The baseline data concerning sleep variables and working memory were visually
inspected as scattergrams (see Appendix 10) but there were no suggestions of
curvilinear relationships. Regression analyses confirmed that no significant quadratic
relationships existed between verbal or visual working memory and any of the sleep
variables measured:
diary time in bed
diary total time asleep
diary sleepiness
actigraphy time in bed
actigraphy total time asleep
actigraphy sleep efficiency
80
Overall, the relationship between sleep variables and working memory at baseline in
the current sample was not very strong. The diary variables were not significantly
correlated with visual working memory, and there was a small but significant negative
correlation with verbal working memory (shorter time in bed and shorter sleep were
related to higher verbal working memory scores). Including socioeconomic status as a
control variable resulted in no significant correlations between any sleep diary and
working memory variables. Actigraphy variables were not significantly correlated with
visual or verbal WM, although controlling for SES showed a weak positive relationship
between efficiency and visual WM.
Measurement issues
Comparison of diary data with actigraphy data
At Time 1, a total of 37 children returned both diary and actigraphy data. Mean time in
bed and mean time asleep for weeknights (Monday to Thursday inclusive) was
calculated in minutes for each of these children.
Table 14: Descriptive statistics for sleep diary and actigraphy data at Time 1
Mean time in bed
Standard deviation
Mean total time asleep
Standard deviation
Diary 10 hrs 21 mins 40.5 mins 9 hrs 39 mins 34.8 mins
Actigraphy 9 hrs 27 mins 37.5 mins 8 hrs 8 mins 41.8 mins
Both diary and actigraphy measures of time in bed and time asleep were normally
distributed, so Pearson’s correlation coefficients were calculated (Table 15). Diary
time in bed was significantly correlated with both actigraphy measures, although the
confidence intervals are wide and therefore the data do not show a robust relationship.
81
Table 15: Correlations between diary and actigraphy measures at Time 1
Time in bed (diary)
Total time asleep (diary)
Time in bed (actigraphy) r = 0.333 p = 0.044 95% CI r = -0.027 to 0.661
r = 0.243 p = 0.148 95% CI r = -0.075 to 0.535
Total time asleep (actigraphy)
r = 0.359 p = 0.029 95% CI r = 0.098 to 0.580
r = 0.308 p = 0.063 95% CI r = 0.059 to 0.567
Paired-samples t-tests showed that the differences between diary and actigraphy
measures of time in bed (t = 7.269, df = 36, p < 0.001) and total time asleep (t =
12.077, df = 36, p < 0.001) were statistically significant, with diary data estimating more
time than actigraphy.
The individual night for which most children (n = 36) returned both diary and actigraphy
data was Monday at Time 1, so this night was used to compare the times at which
children fell asleep and woke up, by diary or actigraphy measure. One child’s data for
time of waking had to be excluded from the analysis as their diary reported waking at
10.10 am, suggesting that they did not attend school that day. Times of day were
transformed using Excel, into the proportion of 24 hours elapsed since the previous
midday, so that each time was represented as a number which could be analysed in
SPSS. As some of these data significantly departed from the normal distribution
according to Kolmogorov-Smirnoff tests (actigraphy time fell asleep: D(35) = 0.155, p =
0.032; diary time woke up: D(35) = 0.155, p = 0.033), Wilcoxon Signed Ranks tests
were conducted. The two samples were not significantly different from each other,
either in the time the children fell asleep (T = 372, p = 0.54) or the time they woke up (T
= 224, p = 0.313). However, because this is a rank order statistical test, it does not
compare means and therefore systematic differences would not be detected (e.g. if
children fell asleep in the same order by both diary and actigraphy measures, but later
by one measure than the other). The mean, median and modal times for falling asleep
82
and waking up were therefore calculated using Excel (Table 16), and suggest that
diary and actigraphy measures are indeed in close agreement.
Table 16: Times of falling asleep and waking up at Time 1
Diary
Actigraphy
Mean time of falling asleep
9.38 pm 9.23 pm
Median time of falling asleep 9.30 pm 9.40 pm Modal time of falling asleep 10.00 pm 9.40 pm Mean time of waking up
7.22 am 7.19 am
Median time of waking up
7.20 am 7.18 am
Modal time of waking up
7.00 am 7.27 am
Differences in the amount of time spent asleep on the two measures do occur from the
amount of waking during the night. Twenty-nine children reported in their diaries that
they did not wake at all during the Monday night of Time 1, five children reported one
waking, and two children reported more than one waking. By contrast, the actigraphy
data showed that these children experienced between 12 and 44 episodes of waking
during that same night (Wilcoxon Signed Ranks T = 666, p < 0.001). In terms of total
time (wake after sleep onset or WASO), the range reported in children’s diaries was 0
to 50 minutes, while the range recorded from actigraphy was 18 minutes to 2 hours 36
minutes, with a mean of 1 hour 1 minute (Wilcoxon Signed Ranks T = 666, p < 0.001).
Compared to the actigraphy data, the children systematically overestimated both the
time they had spent in bed and the total time asleep when they completed their diaries.
The correlations between diary and actigraphy measures of time in bed were
statistically significant, but with wide confidence intervals. These results suggest that
the diary and the actigraphy data are measuring different phenomena, so diary data
cannot be substituted for objective measures of time asleep.
83
Wearing an actigraph did not affect diary data
A total of 71 children returned sleep diaries at both Time 0 (before they knew who
would get an Actiwatch) and at Time 1. At Time 1, 29 of these children were wearing
an Actiwatch and 43 were not. The mean time asleep reported by diary (Monday –
Thursday nights inclusive) was calculated for both time points.
Kolmogorov-Smirnoff tests showed that the difference (in minutes) between diary-
reported time asleep at Time 0 and Time 1 was normally distributed for the group who
got an actigraph (D(29) = 0.096, p > 0.2), but not for the group who remained without
an actigraph (D(43) = 0.153, p = 0.013). One child reported an extremely late night
during the week of Time 0 which skewed the data, and removing this outlier normalised
the distribution of children in the no-actigraph group (D(42) = 0.107, p > 0.2). The
comparison was therefore made (excluding the outlier) using a t-test. This analysis
showed that the children who got actigraphs did not change their diary reporting any
more than those who did not get actigraphs (t = -0.278, p = 0.782). The means for
both groups were small (no-actigraph mean = -3.17 minutes, SD = 29.8 minutes;
actigraph mean = -1.1 minutes, SD = 32.1 minutes), which hides a variety of
differences in both directions: some children reported longer sleep at Time 0 while
others reported longer sleep at Time 1.
These findings suggest that future researchers do not need to be concerned that
wearing an actigraph will affect how children fill in their sleep diaries. However, the
analysis does highlight relatively wide variation between children’s reports of their
sleep duration from one week to the next: even removing the outlier, the range fell
between -96 minutes and +86 minutes’ difference in mean sleep duration).
84
Intervention effects
Sleep knowledge
At the first data collection point, 209 children completed the questionnaire, and, “don’t
know,” responses were coded as incorrect (as per Benveniste et al. (2012). Ceiling
effects were less of a problem with the current sample (see Table 17). The mean was
significantly lower (mean = 4.81, SD = 1.63) compared to Benveniste’s study (mean =
6.36, SD = 1.8) (t = 11.065, df = 850, p <0.0001). Only two items attracted 60% or
higher correct responses, compared to seven items in Benveniste et al. (2012).
Table 17: Responses to sleep knowledge questionnaire at Time 1
Question % correct Benveniste et al. (2012)
% correct Current study
1. Sleep helps me remember things
69.2 54.1
2. Boys have deeper sleep than girls
23.1 37.8
3. Doing something calm before bed can help me fall asleep
84.6 78.5
4. I should go to bed at the same time every night
72.3 43.5
5. Exercising every day can help me sleep better
70.8 49.3
6. I need about 7 hours’ sleep each night (was, “how many hours should you sleep every night to be healthy?”)
69.2 34.9
7. Watching TV before bed can make it hard for people to sleep
60.0 53.6
8. Not getting enough sleep can make people overweight
24.6 15.3
9. Doing exercise just before bed can help me fall asleep
32.3 40.7
10. Children need less sleep than their parents each night (was, “I need more than 9 hours of sleep each night”)
75.4 73.2
85
In the current study, the sleep knowledge data were not normally distributed in either
the control or the intervention group, at any of the three time points. Visual inspection
of the histograms showed that the data were skewed in opposite directions in the
control compared to the intervention group. In addition, the intervention group scored
significantly lower than the control group at Time 1. Despite this difference at baseline,
the intervention group had caught up with the control group at Time 2 (straight after the
intervention), and by Time 3 (8-12 weeks after the intervention finished) the
intervention group maintained higher scores while the control group fell back to
baseline levels (see Table 18 and Figure 6).
Table 18: Effects of intervention on sleep knowledge
Time 1 knowledge scores
Time 2 knowledge scores
Time 3 knowledge scores
Control group mean = 5.06 SD = 1.65 Mean rank = 114.59
mean = 5.82 SD = 1.995 Mean rank = 98.07
mean = 5.02 SD = 1.85 Mean rank = 93.01
Intervention group
mean = 4.55 SD = 1.57 Mean rank = 94.94
mean = 6.23 SD = 1.48 Mean rank = 114.80
mean = 5.72 SD = 1.90 Mean rank = 114.84
Mann-Whitney U test of difference between groups
U = 4431 n = 209 p = 0.017
U = 6430.5 n = 211 p = 0.043
U = 6419 n = 208 p = 0.008
86
Figure 6: Sleep knowledge at Times 1, 2 and 3
Sleep behaviours
The lack of diary data from the intervention group makes it impossible to analyse
whether the intervention group differed significantly from the control group in self-
reported measures of sleep. Only 20 children in the control group and seven children
in the intervention group returned diary data for all time points. Even if we looked only
at the immediate pre- and post-intervention data points (Times 1 and 2), 46 children in
the control group and nine in the intervention group submitted diaries at both points.
Dropout rates were high, with 40% of the control group and 84% of the intervention
group who had submitted an initial diary failing to submit a diary at Time 1, Time 2 or
both.
87
Table 19: Descriptive statistics for actigraphy data at Times 1, 2 and 3
Time 1 Time 2 Time 3
Time in bed (minutes)
Control: n = 37 mean = 573.95 SD = 34.37 Intervention: n = 37 mean = 557.05 SD =29.83
Control: n = 29 mean = 554.72 SD = 48.07 Intervention: n = 25 mean = 558.34 SD =34.73
Control: n = 31 mean = 562.84 SD = 31.70 Intervention: n = 23 mean = 563.20 SD = 30.32
Total time asleep (minutes)
Control: n = 37 mean = 493.36 SD = 32.46 Intervention: n = 37 mean = 485.40 SD =41.83
Control: n = 29 mean = 474.34 SD = 42.42 Intervention: n = 25 mean = 496.56 SD = 30.57
Control: n = 31 mean = 491.31 SD = 47.71 Intervention: n = 23 mean = 488.53 SD = 48.84
Sleep efficiency
Control: n = 37 mean = 86.1% SD = 3.6% Intervention: n = 37 mean = 87.1% SD =5.7%
Control: n = 29 mean = 85.5% SD = 4.1% Intervention: n = 25 mean = 89.0% SD =3.9%
Control: n = 31 mean = 88.2% SD = 4.4% Intervention: n = 23 mean = 89.0% SD = 4.6%
The actigraphy outcome measures were total time in bed (mean minutes per
weeknight), total time asleep (mean minutes per weeknight) and sleep efficiency
(mean percentage per weeknight). All of these measures, at all time points, were
normally distributed according to Kolmogorov-Smirnov tests, enabling the use of
ANOVA.
Looking first at pre- and post-intervention data (Time 1 and Time 2), data were
returned from 28 children in the control group and 25 in the intervention group. There
is a clear interaction between time point and group for total time asleep as measured
by actigraphy (F(1,51)= 4.675, p = 0.035, partial η2 = 0.084). It should be noted that
this interaction was contributed to by those in the intervention group increasing their
total time asleep and those in the control group decreasing it; the interaction does not
represent an effect of the intervention compared to no change in the control group.
88
Focusing just on the intervention group, the amount of sleep increase between Time 1
and Time 2 was not as great as in Sadeh’s (2003) study, with a mean extension in this
study of 11.16 minutes’ total time asleep (Cohen’s d = 0.3; Sadeh et al. (2003)
achieved a mean extension of 35 minutes’ time asleep, Cohen’s d = 0.7). A paired-
samples t-test showed that the difference between mean total time asleep at Time 1
and Time 2 for those in the intervention group was not significant (t = -0.970, df = 24, p
= 0.342). Counting the individuals involved, 14 children reduced their total time asleep
at Time 2 compared to Time 1 and 11 children increased their total time asleep. These
findings suggest that the intervention did not systematically increase sleep duration.
Staying with analysis of pre- and post-intervention data only, interactions between time
point and group were also found for sleep efficiency (F(1,51) = 4.091, p = 0.048, partial
η2 = 0.074) but not for total time spent in bed (F(1,51) = 2.160, p = 0.148, partial η2 =
0.041). Again, the control group reduced their mean sleep efficiency at Time 2 while
the intervention group increased theirs, so the interaction reflects both changes rather
than just an effect of the intervention.
When the time asleep data from Time 3 were added in (n = 25 in the control group and
n = 20 in the intervention group), both groups returned to their baseline sleep
durations, and the overall interaction between time point and control/intervention group
became non-significant in the ANOVA analysis (F(2,42) = 2.339, p = 0.11, partial η2 =
0.052) (see Figure 7).
Likewise, adding in data from Time 3 to the sleep efficiency analysis rendered the
interaction between time point and group non-significant (F(2,42) = 1.465, p = 0.238,
partial η2 = 0.033). In this case, the intervention group increased their sleep efficiency
between Times 1 and 2 and sustained this improvement, while the control group
increased their sleep efficiency between Times 2 and 3 (see Figure 8).
89
Figure 7: Mean total time asleep (in minutes) as measured by actigraphy at Times 1, 2 and 3
Figure 8: Mean sleep efficiency as measured by actigraphy at Times 1, 2 and 3
90
Overall, these analyses suggest that the intervention used in this study generated a
mean extension of 11 minutes in weeknight sleep duration as measured by actigraphy.
The effect was not statistically significant, and more individuals in the intervention
group decreased their sleep duration than increased it. This small increase in mean
sleep duration was not sustained over time, unlike the gains made by the intervention
group in sleep knowledge. Sleep efficiency was also increased in the intervention
group compared to the control group, but at follow-up the control group had also
increased their sleep efficiency so this change may not be an effect of the intervention
itself.
Working memory outcomes
The initial measures of working memory (Time 1) showed that the control and
intervention groups differed significantly at baseline. Both verbal and visual working
memory scores were not normally distributed, so non-parametric Mann-Whitney U
tests were used to compare the control and intervention groups. For verbal working
memory the intervention group started with higher scores (U(215) = 6567, z = 1.751, p
= 0.08, r = 0.12), whereas for visual working memory the control group started with
higher scores (U(215) = 1641, z = -2.490, p = 0.017, r = 0.17).
For both verbal and visual working memory, the control and intervention groups
improved their scores across time points as would be expected with practice and
maturation effects. In both cases, the group with the lower scores caught up to the
other group (see Figure 9 and Figure 10).
91
Figure 9: Verbal WM raw scores at Times 1, 2 and 3
Figure 10: Visual WM raw scores at Times 1, 2 and 3
92
New variables were calculated for visual WM, verbal WM and actigraphic total time
asleep, representing the change scores from Time 1 to 2, from Time 2 to 3 and from
Time 1 to 3. Most of these variables were not normally distributed according to
Kolmogorov-Smirnoff tests (Table 20), so non-parametric analyses were used.
Table 20: Tests of normality for change scores in sleep diary and WM variables
D
d.f. p
Verbal WM change from Time 1 to Time 2
0.115 210 < 0.001
Verbal WM change from Time 2 to Time 3
0.09 202 < 0.001
Verbal WM change from Time 1 to Time 3
0.11 203 < 0.001
Visual WM change from Time 1 to Time 2
0.087 210 0.001
Visual WM change from Time 2 to Time 3
0.101 202 < 0.001
Visual WM change from Time 1 to Time 3
0.076 203 0.006
Actigraphy time asleep change from Time 1 to Time 2
0.099 53 > 0.2
Actigraphy time asleep change from Time 2 to Time 3
0.14 46 0.025
Actigraphy time asleep change from Time 1 to Time 3
0.111 53 0.127
Mann-Whitney U tests of the changes in WM scores between each time point (Table
21) showed that the only significant differences between groups was for verbal WM,
where the control group made greater gains than the intervention group between
Times 1 and 2, and for visual working memory, where the intervention group made
greater gains than the control group between Times 1 and 3.
93
Table 21: Mann-Whitney U tests of changes in WM scores between control and intervention groups
Change from Time 1 to Time 2 (n = 210)
Change from Time 2 to Time 3 ( n = 202)
Change from Time 1 to Time 3 (n = 203)
Verbal WM U = 3951 z = -3.548 p < 0.001 r = -0.24
U = 5684.5 z = 1.418 p = 0.156 r = 0.10
U = 4507.5 Z = -1.537 p = 0.124 r = -0.11
Visual WM U = 6080 Z = 1.314 p = 0.189 r = 0.09
U = 5530 Z = 1.044 p = 0.297 r = 0.07
U = 6231 Z = 2.598 p = 0.009 r = 0.18
Pearson’s correlations with bootstrapped confidence intervals were calculated to
test whether changes in sleep duration, regardless of the intervention, were related
to changes in working memory performance (Table 22). No significant correlations
were found between the change in actigraphic time spent asleep and the change in
WM scores. One correlation was found between the change in diary time spent
asleep and the change in visual WM scores, but the confidence interval was wide
which casts doubt on its significance.
94
Table 22: Correlations between changes in total time asleep and changes in WM scores
Verbal WM
Visual WM
Diary Time 1 to Time 2 (n = 54)
r = 0.139 CI = -0.148 to 0.397
r = -0.116 CI = -0.312 to 0.095
Diary Time 2 to Time 3 (n = 48)
r = -0.148 CI = -0.430 to 0.145
r = -0.097 CI = -0.344 to 0.159
Diary Time 1 to Time 3 (n = 61)
r = 0.002 CI = -0.126 to 0.485
r = -0.228 CI = -0.437 to -0.117
Actigraphy Time 1 to Time 2 (n = 51)
r = -0.039 CI = -0.183 to 0.217
r = 0.199 CI = -0.126 to 0.485
Actigraphy Time 2 to Time 3 (n = 44)
r = 0.216 CI = -0.289 to 0.552
r = -0.035 CI = -0.428 to 0.352
Actigraphy Time 1 to Time 3 (n = 53)
r = 0.046 CI = -0.193 to 0.273
r = -0.179 CI = -0.501 to 0.162
Overall it is not possible to conclude that this intervention has had any reliable
impact on working memory performance. The improvements in both groups suggest
that practice and maturation effects had more influence on WM scores than the
intervention. Differences between the groups are small, and could be explained by
regression to the mean or random effects given that the intervention group did not
always show greater improvement than the control group. There is also a lack of
correlation between change in sleep duration and change in WM scores, suggesting
that those children who did increase their length of time asleep showed no
systematic change in WM performance. These findings are perhaps not surprising,
given the weak associations between sleep and WM variables at baseline.
Programme acceptability
The four classes who had undertaken the intervention were asked for feedback about
their experience of using the modified ACES programme. Teachers and students were
95
asked to complete a brief questionnaire in class as part of the Time 3 data collection,
and questionnaires for parents were sent home with the children.
In total 98 children, four teachers and 17 parents returned questionnaires. All of the
children and teachers who were present on the day of whole-class data collection
completed a questionnaire. The parent questionnaires were largely returned from one
class (n = 14), with the other classes returning two, one, and zero parent
questionnaires. The feedback questionnaires were anonymous in order to encourage
open and honest responses, however this approach does preclude any analysis
comparing feedback to other individual data.
The four teachers were unanimous in their views that the programme had been
enjoyable, practicable, appropriate for the age range and useful. They commented
that it had been worthwhile to raise the profile of sleep as a health behaviour, along
with exercise and healthy eating. Their suggestions for improvement all involved
additions to the programme: those who had not implemented the project said that they
would do so next time, and those who had done the project thought that they would do
more activities with the whole class and involve parents more directly.
The sample of parents returning questionnaires is relatively small, and skewed towards
one class. Of those 17 parents, all reported that their child learned something and 13
(76%) thought that they had learned something themselves. In terms of behaviour
change, 13 parents (76%) reported that their child’s sleep had changed as a result of
these lessons. Three parents wrote an additional comment, all to say that they thought
the programme was useful.
Among the children, the feedback was generally positive, and chi-squared tests
confirmed that there were no significant differences between classes. Most children
96
enjoyed the programme, thought that sleep knowledge and habits were important, and
said that they had changed their sleep habits as a result of the intervention. In Table
23, the numbers of responses indicating positive feedback have been highlighted by
shaded boxes and summarised in the right-hand column.
Table 23: Feedback from children about the intervention programme
Yes, definitely
Yes, a bit
No, not really
No, definitely
not
Positive feedback
The lessons were boring
12 20 41 25 66 (67.3%)
I enjoyed learning about sleep
32 43 15 8 75 (76.5%)
I think other kids should learn about sleep too
72 18 6 2 90 (91.8%)
The programme was a waste of time
4 18 25 51 76 (77.6%)
I think it’s important to have good sleep habits
63 20 10 5 83 (84.7%)
I have changed my sleep because of these lessons
18 42 20 18 60 (61.2%)
Of the 98 children completing feedback, 76 wrote a comment about what they had
most enjoyed as well as responding to the multiple choice questions. Fifty-three of
these comments (70%) indicated that the best thing about the programme was some
part of the assessment procedure (TEACh, n = 14; d2 test, n = 6; Mister X, n = 5; Digits
Backward, n = 1; sleep knowledge quiz, n = 1; computer tests, n = 5; tests in general, n
97
= 5; sleep diaries, n = 8; Actiwatch, n = 8). The remaining comments indicated that the
best thing was the project (n = 10) or learning about sleep (n = 13).
Twenty-two children also wrote a comment about something they would change if the
programme was done with other children. Most suggested making the programme
more fun (n = 7) or including more activities (n= 6). Specific suggestions included
more group work and more pictures in the materials. Two children suggested that it
would be better to have Actiwatches for everyone in the class, and two suggested a
more exciting project. The other comments related to changing the workbook or diary
sheets (without specific suggestions as to what should be changed) and including
more facts in the lessons.
98
Discussion
Key findings
The original overarching question for this research was, “what are the effects of sleep
duration on working memory in children aged 9-10 years?” From this study, the
general answer to this question would be that no effects of sleep on working memory
were found, although difficulties with the research design and implementation may well
have obscured any relationship that does exist.
Within this general research question, specific questions were formulated. The first of
these was whether sleepiness, sleep duration or efficiency correlated with working
memory performance. The evidence for this question comes from the baseline data,
which is relatively good quality information that does not suffer from the dropout and
group selection problems encountered with the intervention part of this research.
Sleepiness did not correlate with verbal or visual memory scores in any analysis.
Sleep duration did correlate with verbal working memory according to diary data (but
not actigraphy data), however this relationship became non-significant when
controlled for socioeconomic status. Sleep efficiency showed a weak relationship
with visual working memory, which became significant when controlled for
socioeconomic status. Overall, the evidence for correlations between sleep and
working memory variables from this study is not strong but suggests that shorter
sleep may be related to better verbal working memory while greater sleep efficiency
is associated with better visual working memory.
These cross-sectional findings would be most consistent with those of Geiger et al.
(2010) and Van der Heijden et al. (2013). The idea that shorter sleep should be
associated with better working memory performance is counter-intuitive and different
99
from the findings in adult studies of sleep deprivation. Geiger et al. (2010) suggest
that one hypothesis to explain this pattern is neural efficiency theory (Haier et al.,
1988), which posits that people with higher cognitive efficiency can achieve
cognitive tasks with lower neural effort, so by extension perhaps those scoring
higher on cognitive tests need less sleep. If this were the case we should expect to
see the same pattern in adults, however no correlational studies of normal
interindividual differences could be found. Another possible explanation given by
Geiger et al. (2010) is that children who sleep for a shorter period of time benefit
from a longer period of wakefulness, and the extra stimulation they gain from this
could accelerate their cognitive development. This hypothesis, although lacking in
empirical evidence, would fit with a developmental view of the relationship between
sleep and cognition, as prolonged wakefulness might be more beneficial in
childhood than in later years when skill development has plateaued.
A further specific research question was whether sleep education improves
sleepiness, sleep duration or efficiency. In this study, the intervention was relatively
brief (three lessons) and only half of the intervention classes did the project intended
to consolidate their learning. Parents were involved through a leaflet sent home.
Assessment of sleep knowledge showed that the intervention did produce an
increase in children’s scores, which was sustained at follow-up, so they had learned
and retained information about sleep.
Not enough diaries were returned after the intervention to assess change in self-
reported sleep behaviours. Actigraphy data did appear to show a significant
difference in sleep duration between the control and intervention groups using
repeated-measures ANOVA, however this difference was due to the control group
decreasing their mean sleep duration as well as the intervention group increasing
theirs. No significant increase in sleep duration was found for the intervention
100
group. Taking into account pre-intervention, post-intervention and follow-up
actigraphy data, the intervention group did not increase their sleep duration or
efficiency any more than the control group. These findings are consistent with those
of other studies of sleep education programmes, which tend to impact upon
knowledge but not behaviour (Blunden et al., 2012).
There could be a number of explanations for this lack of effect on sleep behaviours.
First, most children were getting enough sleep at baseline, compared to normative
data on reported sleep durations (Matricciani et al., 2012). These children would
therefore not be expected to change their sleep habits on the basis of education
which would inform them that their sleep duration was about right. It would have
been useful to analyse just those children with short sleep durations at baseline to
see if the intervention changed their sleep behaviours, but not enough data were
collected in this study to do that. Second, the intervention may not have been
intensive enough to produce behaviour change. Third, more active parental
involvement may be required in order to make changes to routines occurring at
home. Finally, there may be factors preventing behaviour change even if the child
intended to do so, such as perceived importance, locus of control and practical
issues such as room sharing or night-time activity of others in the house.
The question of whether changes in sleep correlate with changes in working
memory performance cannot be answered by the current study, as the intervention
was not successful in altering sleep variables.
The acceptability of the intervention was also assessed. Most children, staff and
parents who responded were positive about the programme. Ratings were similar to
other sleep education programmes (summarised by Blunden et al., 2012), with over
90% of children in the current study saying that other kids should learn about sleep
101
but only three-quarters enjoying the programme. From looking at other evaluations
covered in the review of Blunden et al. (2012), ratings of enjoyment tend to be
higher when the programme is delivered by an external professional rather than the
regular teacher.
This study was able to answer its methodological questions using baseline data.
First, the question of whether sleep diaries reflect actigraphy data in this age group
was resolved with findings which are consistent with other studies of child
participants (e.g. Holley et al., 2009). There was a systematic over-estimation of
sleep duration in diaries compared to actigraphy, which is mainly attributable to a
lack of reporting wake after sleep onset (WASO) and a possible over-estimation of
WASO by actigraphy. In general there was agreement between the two methods
about the times children went to bed and woke in the morning.
A novel question which has not previously been addressed was that of whether
wearing an actigraph would affect the way children fill in their sleep diaries. Analysis
of the diary sheet completed in the first week (before children knew who would get
an actigraph) and in the following week (when some children wore an actigraph)
showed that there was no difference in diary reporting between those who did and
did not receive an actigraph. Future studies can omit the baseline diary week used
in this study, as it added no new information to diary data collected at the same time
as actigraphy data.
Key methodological issues
In the current study, allocation of classes to the control or intervention group was not
random. The method of allocation was practical and ethical, as it allowed all classes to
have the intervention (if they so wished) within the academic year and engaged
102
teachers by giving them control over when to deliver the intervention. However, the
keenest teachers wanted to start the project as soon as possible and this introduced
bias into the group allocation. The most highly motivated teachers, in the best
performing schools, with higher SES populations, volunteered to go first and therefore
formed the control group. Random allocation of classes to control and intervention
groups would have given a better chance of more comparable groups. In addition,
recruitment of teachers in the summer term, ready to undertake the project from the
beginning of the new academic year, would have allowed for randomisation of when
each class fitted into the project schedule rather than having all the control classes
first.
Assessment of working memory relied upon only one test each of verbal and visual
working memory. The original study design included three subtests for each, which
would have given a more reliable score, but the chosen assessment tool was
withdrawn just before the start of the project. While other studies have also used the
same single tests of WM as in this research (Cho et al., 2015; Dunning & Holmes,
2014), their validity in this project is questionable. The current sample did show scores
comparable to the standardisation sample, however their scores for the visual and
verbal WM tests were not correlated with one another which is unusual.
The assessments of attention were not used in the data analysis, since no main effects
for working memory were found and therefore no mediator analyses were carried out.
The measure of visual attention was valid and reliable, with the current sample
showing a normal distribution of scores. However, the measure of auditory attention
suffered from ceiling effects, with a skewed distribution and the majority of children
scoring 9 or 10 out of 10. A different measure may have been preferable, however no
other group-administered auditory attention tests could be found and the resources to
test each child individually were not available.
103
The data collection for sleep variables had good participation at first, but children
quickly got bored of filling in sleep diaries and the dropout rate was so high that
longitudinal analysis was not possible. A focus group was held, in which six children
who had not returned sleep diaries were asked to comment on why they had not done
so. This focus group was held after the issue of not returning diaries had become
apparent, and was held with the next class to finish their data collection, so that there
was no pressure on them to complete diaries in the future. The children all agreed to
participate in the focus group and all contributed to the discussion. Their reasons for
not returning sleep diaries were:
Lost the sheet
Spent time away from home and did not take diary with them
Living between two homes and did not take diary to the other home
Spilled a drink on the sheet
Forgot to do it
The children also had ideas for how to increase the return rate in future:
Teachers to check who has brought theirs back and remind those who have
not
Teachers to provide incentives for returning diaries
Remind participants that returning their diaries will help other people
Put information about sleep on the back of the sheet to remind participants why
they are doing the study
Make the sheet more fun and engaging, with pictures and puzzles on the sheet
Treatment fidelity was also an issue. Two out of the four intervention classes
completed the project intended to consolidate and personalise children’s learning; the
104
other two could not find the time to do the project. Even within the lessons, which were
highly manualised, some teachers reported omitting some of the activities. If the
resources had been available, observation of programme delivery would have been
ideal. Alternatively, teachers could have been asked to complete a checklist of
elements of the intervention, to encourage a complete delivery and to identify any
elements not delivered (e.g. Smith, Daunic, & Taylor, 2007). While there is an
argument that, for real-world studies, interventions need to be tailored and allow
practitioners some freedom in their use, this flexible approach may more useful for
programmes which are already well validated so that the research can focus on
variations from the established intervention.
Reflecting on the research design, it may be that universal sleep education
interventions require much longer-term evaluation in order to detect significant impact,
or to be delayed until the prevalence of inadequate sleep is higher. In the 9-10 year
old children participating in this study, three-quarters were already getting the
recommended amount of sleep and therefore did not require intervention. It is possible
that, as they move into adolescence when sleep durations often reduce, those children
who have had sleep education may show some behavioural effects compared to those
who have not. Alternatively, universal sleep interventions may not be effective until a
higher proportion of children are not getting enough sleep and would therefore benefit.
A different approach would be to target sleep interventions only at those children
already experiencing sleep problems, which have so far shown larger effects than
universal programmes (Blunden et al., 2012; Quach et al., 2011).
A further issue for the research design is that of intra-individual variability. While the
current study has focused on habitual sleep behaviours, the variability in the same
individual’s sleep duration on different nights is wide. It is therefore difficult to identify
reliable changes in sleep habits, as the standard deviation is so large at baseline.
105
Perhaps this variability would have been a useful measure in itself, as other studies
have found that consistency in children’s sleep routines is related to outcomes
(Pesonen et al., 2010; Wolfson & Carskadon, 2003).
Finally, there are methodological issues concerning the sample size in the current
study. Although previous research was used to calculate an appropriate sample size,
different studies have found different effect sizes. Selected studies which were in
some way similar to the current research design were used to estimate a sample size
which would have detected significant relationships if they existed in similar ways to the
reference studies. However, the meta-analysis of Astill et al. (2012) found an effect
size for the relationship between children’s sleep duration and executive functions of r
= 0.07. This effect size mirrors that found for the relationship between children’s sleep
duration and academic attainments (r = 0.069; Dewald et al., 2010). If the effect in the
population was really as small as Astill et al. (2012) suggest, a sample size of 1596
would be required and this was not practicable in the current study. Indeed, if the
effect size is that small, it is unlikely to be clinically significant and suggests that sleep
is not a very effective target for intervention if the desired outcome is improved
executive functioning in children.
Key theoretical issues
The neurodevelopmental perspective may be helpful in understanding the differential
effects of sleep on functioning in a range of domains over the lifecourse. The
development of the frontoparietal network subserving working memory in adults is
incomplete in this age group, both in functional connectivity and in axonal myelination,
so we might expect to find that different factors affect WM performance depending on
the maturity of these biological systems.
106
It may also be that children, like adults, have inter-individual differences in their need
for sleep. Rather than espousing a recommended sleep duration for children, perhaps
it would be more relevant to track children’s functioning and suggest changes to their
sleep habits only if they are experiencing daytime sleepiness or other functional
difficulties. Evidence from Konen et al. (2015) suggests that there are functional losses
when children depart from their own usual sleep habits, which would not be detected in
a study looking only at group differences in average sleep durations.
There is also a question about the psychology of behavioural change. While the
intervention used in this study is based upon the theory of planned behaviour (Ajzen &
Fishbein, 1980), there are challenges to the efficacy of this approach to changing
people’s behaviours (Armitage & Conner, 2001; Hardeman et al., 2002). It may be that
sleep education is never going to be an effective way of changing children’s sleep
behaviours; even if their knowledge is improved and they have intentions to change
their behaviours, a range of factors are likely to impede implementation of that change.
For example, sharing a bedroom, parental expectations or night-time activity in the
home (e.g. adults coming home from work or leisure activities after the child’s bed
time) may work against a child acting on their intentions to go to sleep earlier. Wake
times, at least during the week, are unlikely to be amenable to change as children have
to get up in time for a fixed school start. Alternative approaches may be required and
some are already being researched, for example changing the start time of the school
day to allow for longer sleep in the morning (e.g. Owens, Belon, & Moss, 2010). Later
school starts are also being investigated in the UK: the TeenSleep project
(https://educationendowmentfoundation.org.uk/projects/teen-sleep/ retrieved 12
August 2015) based at the University of York is due to report in 2018. The US based
studies are more likely to find positive results as their standard school start times are
usually earlier than in the UK (e.g. Howard-Jones, 2014; Kirby, Maggi, & D'Angiulli,
2011). However, a Norwegian study (Vedaa, West Saxvig, Wilhelmsen-Langeland,
107
Bjorvatn, & Pallesen, 2012) provides some evidence that moving secondary school
start times from 8.30 am, which is more similar to UK start times, to 9.30 am improves
reaction times (but not sleepiness or mood).
Cognitive psychology may also illuminate the theoretical underpinnings of the current
research. Cognitive functions appear to be separable, even in children of this age, and
show specificity in their development (e.g. Casey, Tottenham, Liston, & Durston,
2005). Two relevant arguments result from this understanding. First, if the desired
outcome is an improvement in working memory performance, the most effective way to
achieve that is to train working memory directly. Factors such as sleep, exercise and
diet may set the general framework for cognitive development, but specific outcome
measures will show large changes only if targeted directly for intervention. Second,
development of WM skills may not show generalisation to real-world outcomes of
interest. The literature on specific WM training shows that WM performance can be
improved by sustained and adaptive training, in a range of populations, however such
improvements do not necessarily lead to higher achievements in academic tests
(Melby-Lervag & Hulme, 2013). Taking a cognitive psychology perspective might
suggest that the effects of even the most effective sleep intervention on specific
cognitive functions may be weak, and the effects on real-world outcomes even weaker.
This conclusion does not mean that sleep interventions are not worth doing, but that a
realistic view of their effects within complex human systems would need to be taken.
Future research
There are some implementation aspects of the current study that would strengthen it if
a similar design were to be used again. Programme fidelity could be tracked more
closely and a greater emphasis placed on delivering the programme as set out in the
materials given. There could also be greater focus on increasing participation rates.
108
The current study was able to collect large amounts of data from children in class, but
the measures requiring home completion suffered high dropout rates. The children
themselves gave some useful ideas for increasing participation, such as more
attractive diary sheets, teacher tracking of data returns and rewards for participation.
In addition, technological solutions to data collection might increase participation, for
example mobile phone based prompts to complete diary information as demonstrated
by Konen et al. (2015).
Recruitment of a targeted sample could be adopted for future research, to give a
greater chance of changing sleep behaviours. For example, initial assessment could
be used to target children who are short sleepers or who show problems associated
with sleep such as daytime sleepiness. This targeted approach would enable most
children to be excluded from the study as they do not need the intervention, and allow
a focus on those children who are most likely to benefit from a change in sleep
behaviours.
The intervention itself may also need to be adapted in order to produce changes in
sleep behaviour. A longer programme may be needed, perhaps with learning spaced
over time in order to maintain any changes in behaviour. Parental involvement may
need to be more active, particularly in this pre-adolescent age group, in order to
support children in changing sleep habits. Alternatively, it may be that a sleep
education programme is not sufficient to produce significant changes in behaviour and
the intervention needs to have a different focus altogether.
Future research may need to use research designs with multiple outcomes in order to
detect the holistic effects of sleep on daytime functioning. If this study had also
included measures of other functions, more significant results may have been
observed. The existing body of evidence suggests that sleep may have an impact on
109
a range of outcomes including learning, episodic memory, reaction time, obesity and
emotional regulation. It would be unhelpful to conclude that children’s sleep duration is
not important just because it shows no relationship with a single cognitive domain.
Intra- and inter-individual differences in sleep habits will also need to be accounted for
in future work. Tracking of changes in sleep behaviours from night to night, weeknight
to weekend and school term to school holiday periods would be helpful in order to
establish the normal range of variability in children’s sleep behaviours. It would then
be possible to investigate whether variability itself is an important predictor of daytime
functioning and therefore whether it would be a plausible target for intervention.
Individual differences in sleep need may also be important, with the effects of sleep
deprivation or extension being found only in relation to an individual’s usual sleep
pattern.
Future studies could elucidate the developmental aspect of the relationship between
sleep and WM, by choosing a methodology and applying it to different age groups in
order to track the course of the relationship between sleep and WM. Imaging could be
part of the research design in order to link directly the development of the fronto-
parietal network with the effects of sleep on WM. A useful hypothesis to be tested
would be that the effects of sleep on WM get stronger as the neural systems
subserving these functions approach maturity.
110
Critical appraisal
111
The process of research development
The current thesis was my original research idea, however I also pursued a number
of other topics before committing to the project as set out here. The process of
exploring research topics is discussed here, as it provides important context for the
decisions taken in the current study.
From the start of the doctorate I had some general principles in mind for my thesis,
which were that it should further my career aspirations in the field of
neuropsychology and that the project should involve some kind of intervention likely
to benefit the participants (see section on intervention-led research). I had an early
interest in both sleep and working memory, so this was my initial area of focus.
Very early in the research process I considered methods, and from reading the
literature about sleep I was convinced that objective measures were important, as
so many studies showed differences between self- or parent-report and actigraphy
or polysomnography. There was therefore a practical issue of whether I would be
able to access the research tools required: if I was not able to get actigraphy
devices then I would not be doing a project that measured sleep behaviours. I
contacted a range of colleagues in academic, hospital and charity organisations,
who confirmed that actigraphs were practical to use but expensive, and they were
not able to lend theirs out. At that point, I abandoned the idea of studying sleep as I
thought that report methods alone would not give the rigour I wanted to achieve.
At that time, I attended a conference which included an interesting workshop on
traumatic brain injury and offending behaviour. I thought that this specific field of
research was well covered by the speaker and other teams mentioned in his
112
presentation, but further reading in the field of aggressive offending behaviour led
me to Essi Viding’s research on callous/unemotional traits in children. I read an
article (Dadds & Rhodes, 2008) outlining what neuroscience had to offer in terms of
understanding callous/unemotional traits and developing interventions specific to
them. As Essi Viding was working in the same department in which I was enrolled
as a doctoral student, and she had already collaborated with Norah Frederickson in
the educational psychology group, I thought that this could be an opportunity to work
with colleagues at the forefront of their field and to be part of a cutting-edge
programme of research.
I met several times with Profs Viding and Frederickson, and spent a lot of time
reading in this field. I thought about a number of theories which might be relevant,
particularly those relating to reactivity as I thought this concept might help to explain
the differences between aggressive children with and without callous-unemotional
traits and therefore suggest differentiated interventions. I found it increasingly
difficult to narrow down ideas to a project that would be manageable and worthwhile.
I also experienced some tension between input from different colleagues, as they
were drawing on different perspectives, experiences and expectations in their efforts
to help guide my thinking. A key turning point came when I brought a shortlist of
topics on which I might focus a literature review, and in supervision we crossed out
everything related to neuroscience. It seemed as though the project was becoming
increasingly complex and moving further away from the core ideas I wanted in my
doctoral work: neuropsychology and intervention. A supervision session with my
link supervisor gave a fresh perspective from someone outside the specific field, and
I made the decision not to pursue a thesis on callous-unemotional traits despite all
the work that I and others had put into it.
I got back in touch with Sarah Blunden, a researcher in Australia who had
113
corresponded with me during my initial explorations of a thesis about sleep. She put
me in touch with Teresa Arora at Birmingham University, who investigates the
relationship between sleep and obesity. During a conversation about the resources
needed for doing sleep research, Dr Arora suggested that I approach the
manufacturers of actigraphs directly. I took up her idea, and investigated some
other topics while awaiting their responses. As soon as I heard that Philips were
prepared to loan 10 devices for the 15 months required, free of charge, I committed
to developing a study that would use this resource.
In developing the current research further, I drew on my experiences from the work I
had done on the previous project idea about callous-unemotional traits. I was sure
that my two core principles needed to be represented in the research, as I had
learned that I was not comfortable without neuropsychology and intervention as part
of the programme. I also knew that I had to narrow down the field in order to
generate a manageable project. I felt differently about doing a project where there
were no experts in the field in my department (or, arguably, in the country). This
would mean that my supervision would be more generic and I would have more
control over the content, but perhaps at the expense of having subject-specific ideas
suggested, or knowing the most up-to-date information about ongoing studies.
Research questions, methodology and epistemology
Following the structure put forward by Crotty (1998), a coherent thread is identified
here linking the research question, the methods used, the theoretical perspective
taken and the underlying epistemology.
For the current study, the key question was, “what are the effects of sleep duration
on working memory in children aged 9-10 years?” This question was derived from
114
my own interests, and focused by live practice issues as well as a review of the
literature. I was drawn into the study of sleep during my first degree, when I undertook
a dissertation on narcolepsy and found it fascinating. Postgraduate learning in the field
of neuropsychology has developed further my interest in the physiological and
cognitive aspects of psychology, as layers in our understanding of how people
function. Executive functions have become a particular area of interest, partly due to
their apparent vulnerability in brain injury (e.g. Levin & Hanten, 2005) and partly
because they have been shown to predict educational outcomes (e.g. Blair & Razza,
2007). In a previous job I helped to develop an extensive set of training and resources
to support the assessment and development of executive functions, based on
materials which had themselves been part of someone else’s doctoral work. Feedback
from teachers confirmed that the topic was both novel and useful to them, confirming
that it could be a relevant area for my own research.
Initially, I assumed that there is a relationship between sleep and working memory, and
that the focus of the research would be on the effectiveness of sleep intervention in
changing sleep behaviour and therefore working memory performance. However, as
the literature review proceeded I realised that this relationship was far from clear and
that the next steps in our collective knowledge would be to find out more about whether
sleep duration affects working memory in children at all.
This broad research question led me to a set of methods which would enable some
quantification of both sleep and working memory performance. From the literature
review I was keen to have an objective measure of sleep behaviour, in addition to
sleep diaries kept by participants, and the only practical way to do this without access
to a sleep laboratory was through actigraphy. I contacted the manufacturer of a
commonly used actigraphic device and was pleased to negotiate the free long-term
loan of 10 devices. The availability of this equipment drove the sequential schedule for
115
data collection, as 10 children at a time could be having their sleep behaviours
measured in this way. If no actigraphy kit could have been obtained, I would have
needed to change the research substantially, perhaps moving away from measuring
sleep variables at all. If only one device were available, I might have taken a case-
study based approach and looked at a small number of individuals in greater depth.
This approach would have been idiographic and would have necessitated a
phenomenological perspective. It could have provided information to illuminate the
many complex factors involved in children’s habitual behaviours and cognitive
performance. This sort of research could be helpful in understanding why no clear
relationship between sleep and working memory has yet been established in children.
However, given that larger-scale research was made possible, an experimental
methodology was used, with enough participants to detect correlations in the baseline
data and to form two comparison groups for the intervention.
Experimental methodologies like those used in the current study generally derive from
a critical realist theoretical perspective (see, e.g. Groff, 2004). An objective reality is
assumed, as in positivist views, but in critical realism the researcher accepts that our
data and its interpretation can only approximate that underlying reality. In this way,
research can be conducted with a view to reducing subjectivity and finding results that
can be generalised, while acknowledging that psychological phenomena are unlikely to
be characterised by rules or laws which hold in all contexts for all individuals.
Crotty (1998) suggests that there are three main epistemological positions within which
a theoretical perspective might sit: objectivism, constructionism and subjectivism.
These positions can be viewed in terms of their predictions about the generalisability of
findings. Objectivism suggests that meaningful reality exists independently of any
individual’s experience, and that research should aim to identify aspects of that
objective reality. From this position, rules would always hold so long as the theory
116
matches reality. In contrast, subjectivism suggests that there is no independent reality,
and that meaning is always created by people through their interactions. This position
does not allow for generalisable rules, but focuses on each unique experience.
Between these two extremes lies constructionism, which suggests that meaning is
created through our engagement with a world which has its own properties but can
never be experienced directly. From a constructionist perspective, research can look
to find the extent of generalisability for a particular phenomenon, identifying at least
some of the factors which exert an influence on the rules.
Using Crotty’s (1998) framework, the current research can be summarised as follows:
Figure 11: From research question to epistemology
The current research took a generally objectivist approach. An attempt was made to
identify a general rule about how sleep affects working memory, with a large enough
sample size to uncover some of the objective truth of the matter. Given the difficulties
encountered with children returning sleep data which restricted the sample size, the
generalisability of the findings in this study are severely limited and the research fails to
identify robust relationships between the variables. Even if the research could have
•What are the effects of sleep duration on working memory in children aged 9-10 years? Research question
•Actigraphy
•Sleep diaries
•Standardised testing
•Knowledge questionnaires
•Quantitative statistical analysis
Methods
•Experimental
•Cross-sectional + intervention Methodology
•Critical realism Theoretical perspective
•Objectivism Epistemology
117
been implemented as designed, it would not take into account individual differences in
sleep need or other factors which might contribute to working memory performance.
In hindsight, a constructionist approach may have been more useful. Any relationship
between sleep and working memory is likely to be mediated by a range of factors far
beyond the basic measures in this study (gender and socioeconomic status). A more
fruitful approach from a constructivist perspective might have been to focus on
individuals’ sleep patterns, the reasons for those sleep patterns and a range of
outcomes including cognitive, emotional and physical functioning. In this way, factors
might be identified which mediate not only the relationship between sleep and WM
variables, but also the link between intentions to change sleep behaviour and actual
behaviour change (e.g. bedroom sharing). This kind of research could use qualitative
or case-study based quantitative methodology.
Intervention-led research
A key driving force for undertaking the whole doctoral programme for me has been
to have a positive impact on children and young people. I always wanted to include
an intervention in my major research project, so that the act of carrying out the
research had a chance of benefiting the participants directly.
In addition, as a practitioner, I often find intervention-led research most helpful in
order to inform what might work with the children I serve. Academic research may
be interesting but I am usually left asking, “so what?” This problem is particularly
true in my special interest area of neuropsychology where there are usually many
articles describing a condition and its outcomes but little or no evidence guiding how
a psychologist might help someone with that condition. For example, while working
in a specialist NHS post I met several children whose posterior fossa tumours had
118
been surgically resected. They showed common difficulties, and I read
enthusiastically about posterior fossa syndrome which was new to me. I found
many articles describing the possible sequelae of posterior fossa damage, and
factors which would make these sequelae more or less likely, but not a single article
suggesting how any of these problems could be prevented or reduced.
For my own doctoral research, I was therefore keen to include some element of
intervention in the design. I wanted to be able to say that I had found something
practical which I could recommend, or steer others away from. Through the course
of the thesis, I came to realise that any evidence-based intervention is just the tip of
an iceberg formed of theory, practice and research. In the case of sleep
interventions, the underlying iceberg is probably not yet formed well enough to
support the development of an intervention which stands a good chance of changing
sleep behaviours, and still less chance of changing cognitive abilities. Michie,
Johnston, Francis, Hardeman and Eccles (2008) clearly explain that evidence-based
interventions require a modelling phase which involves, “hypothesising and testing
both what to target (behavioural determinants) and how to do this (techniques to
change these determinants).” (p. 661) In the case of sleep interventions, I think that
we are still very much in this modelling stage. We are not sure yet which sleep
behaviours we need to encourage, how sleep need might change over the course of
development or how individuals might differ in the sleep behaviours which would be
optimal for them. With a clearer idea of such behavioural determinants, or what we
might call, “good sleep,” we would be in a better position to start modelling how to
change those determinants.
In terms of the ethics of participant benefit, I can now see that this project, although
it had an intervention at its core, may not have brought much benefit to the children
involved. The reason for using the sleep education intervention was to generate
119
lasting behaviour change rather than temporary sleep extension or restriction, which
was a laudable aim that was not met. The design used would have been more
robust if an intervention were available that had good evidence of changing sleep
behaviours; given that no such intervention exists yet, a different approach would
have been more appropriate. To answer the research question, it may have been
more effective to use an experimental manipulation as other studies have done (e.g.
Sadeh et al., 2003; Vriend et al., 2013), rather than an intervention with a limited
evidence base which was much less likely to alter sleep behaviour.
Having set out these self-criticisms, I think that the project is unlikely to have harmed
the participants and they have only missed a very small amount of the curriculum
they would otherwise have been taught. There may be long-term benefits in terms
of sleep habits as those children move into adolescence, and there may have been
positive behaviour change not measured by the methods in this study (either due to
missing data or variables which were not assessed, such as other cognitive skills or
emotional regulation). The acceptability data suggests that most participants
thought the programme was enjoyable and worthwhile, so this study is not likely to
have caused negative emotional effects. Looking wider than the participants in this
particular study, although this research was not aimed at evaluating the ACES
programme, it does provide some useful data for practitioners to bear in mind when
considering whether such an intervention might be worthwhile.
Risky research
The study undertaken had many risks associated with its implementation, some of
which occurred and were managed, and others that would have stopped the project
completely. In hindsight, the research design could have been adjusted to reduce
120
some of these risks before starting to implement the project. Table 24 shows the
risks which did happen, and how they were dealt with.
Table 24: Realised risks and their management
Risk
Solution
Attention testing using the CPT took much longer than anticipated
Group-administered test of visual attention was found
Group-administered working memory test was withdrawn from the market
Individual tests were found, along with a volunteer research assistant to administer them
Actiwatch software not compatible with new laptop bought for use in this project
Actiwatches collected on a Friday so they could be uploaded on the desktop and reset
Actiwatches not working correctly One Actiwatch had to be replaced after the pilot group showed it was faulty
Actiwatches not worn correctly Despite verbal and written instructions, some missing data had to be accepted
Actiwatches mislaid Researcher visited several homes to collect Actiwatches not brought into school on collection day. One Actiwatch completely lost – replaced by the supplier at no cost
Sleep diaries not returned Focus group held to find out why and generate
ideas for improving return rates, but it was too late to implement the ideas in this study
Intervention not implemented as directed
Information collected retrospectively about which parts of the intervention were delivered; decision made that all classes had done the core elements so would be included in analysis
If I were to repeat the research, I would place more focus on the teachers, who
could have helped to reduce many of these risks. For example, some teachers kept
track via class lists of which children had returned sleep diaries and reminded those
who had not done so: in future I would explicitly ask all teachers to adopt this
strategy. In addition I would ensure that some measures of treatment fidelity were
121
included. With the time required for data collection it was impossible to visit each
class during delivery of the intervention, however I would at least provide a checklist
of intervention elements that the teacher would complete and return.
Other things could easily have gone wrong during this project, which could have
derailed it entirely, including:
Researcher, research assistant or school unable to stick to timetable (e.g. off
sick)
Laptop or software malfunctioning
Classes withdrawing during the project
More Actiwatches lost or broken
The research schedule was extremely tight, with no spare weeks in case a delay
occurred anywhere in the system. Luckily all the people involved were able to keep
to the schedule every week, so the only missing data was from individual children
who were off school on the day of whole-class data collection or failed to return
sleep data. If I were conducting the same research again, I would aim to use the
whole academic year so that some weeks could be scheduled for no data collection,
thus providing a cushion in case data collection ever needed to be pushed back a
week. In addition to the lack of contingency time, there was no room for error in
collecting and delivering the Actiwatches: failing to do this on time would have
knock-on effects for the next class in the schedule who would be relying on the
devices being transferred to them. A laptop which was compatible with the
Actiwatch software would have helped to solve this problem, as the Actiwatches
could then have been taken straight from school to school on the same day.
Many difficulties in this research were driven by the use of 10 Actiwatches to gather
data from a large group of children. With no funding for this project, and the set of
122
devices being loaned free of charge, using them sequentially seemed to be the best
use of this precious resource. However, it may have been more fruitful to use that
resource differently. It would have been possible to select 10 children, for example,
and track their sleep habits over a longer period of time. This kind of approach
would not have the kind of generalisability gained from larger numbers of
participants, but it might have enabled a detailed analysis of what factors influenced
sleep each night, whether an experimental manipulation changed sleep habits and
whether cognitive variables were related to either long term habits or short term
fluctuations in sleep.
Using advice from supervisors
Looking back, there were a number of pieces of advice from supervisors which I did
not take at the time but which would have improved this research. On reflection, I
think my experiences of working on the callous-unemotional topic perhaps made me
more resistant to accepting supervisors’ advice. The guidance I had received on
that project was different between supervisors, so I felt I needed to trust my own
judgements rather than accepting advice straight away. Unfortunately, in most
cases, their advice was sound in hindsight.
Literature review
A thorough literature review is advised before designing the empirical study. I read
widely and did draft a literature review, however I did not get to the point of selecting
relevant studies to analyse in detail before implementing my data collection. I
designed my study based on authors’ own conclusions about their findings, but
when I went back and carried out my own analysis I realised that the existing studies
showed much less conclusive evidence relating to my specific research question
123
than I had thought. In future I would look at relevant published research in greater
detail, including their methodologies and results, to make a more informed
judgement about the existing state of evidence.
For example, Sadeh et al. (2003) reported that, “the sleep manipulation led to
significant differential effects on neurobehavioural function measures,” (p. 444).
They provide graphs of three cognitive measures which showed different changes
depending on whether the children had extended or restricted their sleep. What I
had not understood until after commencing my own study was that these were the
only three significant results: most of their cognitive measures did not show
significant differences depending on sleep extension or restriction. The authors had
provided the data correctly, but like most reports (Dwan et al., 2008), the article
focuses on significant results rather than null results. The significant relationship
was between sleep duration and digits forward (usually considered a test of short
term memory), whereas the relationship between sleep duration and digits backward
(seen as a test of working memory) was non-significant. In addition the authors
provided a table indicating which interactions were significant in an analysis of
variance. This table included which cognitive measures showed change between
the first and second test, as well as the interaction between time point and group. I
had not appreciated that the interaction was the important statistic (three out of nine
cognitive measures showed significant interactions, and these were the ones
presented in graph form). I saw that the change between the first and second times
of testing was significant for digits backward, and misinterpreted this result as
meaning that sleep had affected this working memory measure. In fact this result
only meant that all the children did significantly better when tested a second time.
124
I have learned a great deal about reading the literature thoroughly and analytically,
which I will be using in future to draw my own conclusions about the existing
evidence before designing research based upon it.
Dependence on others
Another good piece of advice was to design research which depends as little as
possible on others. As the researcher, you are highly motivated to do what needs to
be done, to persevere and to do things as intended. The more you rely on others to
facilitate your research (e.g. carrying out data collection or implementing an
intervention), the greater the chances that the research does not go as planned.
This difficulty is rarely pointed out in standard texts on research design (e.g. Brewer,
2000; Robson, 2011).
The current study was highly dependent on others for its implementation, even more
so with the introduction of a research assistant to help with data collection which
was necessitated by the unavailability of group-administered working memory
measures. Some of this dependency was intentional, for example I wanted to know
if the intervention would work if regular teachers delivered it, rather than an external
professional. However, given that the intervention did not have good evidence of
effecting behaviour change even when delivered by a sleep expert (Blunden et al.,
2013), this was not a good decision in terms of answering my main research
question. Given the scale of this project, some of its dependency on others was
purely practical. There was no funding to buy resources or pay staff, so I was reliant
on the manufacturer to provide the actigraphy devices and to replace faulty or lost
ones (which they did admirably).
125
If I were doing this research again, I would look for ways of reducing reliance on
others. For example, I could have omitted the intervention altogether, or with more
resources delivered it myself. Unfortunately studies of sleep will always require
some co-operation from home, as the data cannot be directly collected in school. It
would have been possible to use a survey to gauge sleep habits, but this
questionnaire approach has been shown to be inaccurate compared with diary or
actigraphy methods (Werner, Molinari, Guyer, & Jenni, 2008).
Where it is necessary to rely on others, I would in future place much more emphasis
on ensuring they knew what was expected and prompting them to undertake their
part. For example, I might have met with each teacher just before they were due to
begin intervention, to provide refresher training and ensure that they were confident
about delivering all aspects of the intervention.
Data input and analysis
I was advised to use SPSS software from the outset, and to generate one master
dataset from which all analyses could be done. However, as I was not very familiar
with that programme at the time, I decided to use a Microsoft Excel spreadsheet
instead. This decision had advantages, for example it made the set-up and data
input quicker because I knew how to use the software. It was also available on all
the computers being used for this project, whereas SPSS was only available on one
laptop. Initial data entry using Excel was not a problem, as data could easily be
imported into SPSS directly.
I then decided to set up different SPSS datasets to answer each of my research
questions. I did this because each topic needed only a subset of the data, so it
seemed more manageable to have small and specific datasets and results pages.
126
In hindsight this was a mistake. First, the data entry is likely to be flawed, and when
a mistake is spotted, having a single dataset would ensure that all future analyses
are carried out with the corrected data. The way that I managed my data meant that
any errors had to be corrected on the Excel spreadsheet and then on any SPSS
datasets which included that datapoint. I did later create an SPSS dataset with all of
the key variables included, with which I double-checked my results from the smaller
datasets and continued with the analysis. Second, as the analysis progressed I
found I wanted to carry out statistical tests involving other variables which I had not
initially copied into SPSS. Third, SPSS deals with calculations from empty cells as
missing data, whereas Excel treats them as zero values. I realised this after I had
calculated variables in Excel such as mean sleep durations, where the means were
lowered if a child had not filled in their sleep time for one night in their diary because
Excel treated this as if they had got no sleep that night. It took some time to check
through all the data and ensure that the calculated variables were all correct. This
problem would have been avoided by using SPSS exclusively.
The (un)importance of sleep?
One reflection from both the literature review and the empirical study is that my
assumptions about the importance of sleep have been challenged. At the outset of
this work, I believed that sleep was a cornerstone of human performance, and with
the advent of an, “always-on,” culture (e.g. Park, 2013), we were all likely to be
sleep deprived and suffering cognitive decrements as a result. However, recent
systematic reviews suggest that over the last 40 years, adults in Britain have
actually increased their average sleep duration (Bin, Marshall, & Glozier, 2012) and
over the last 100 years, British children have bucked the worldwide trend, also
increasing their average sleep duration (Matricciani, Olds, & Petkov, 2012).
Perhaps children here in the UK are not sleep deprived after all, on a population
127
level. The literature demonstrating the cognitive impact of lack of sleep in adults is
largely based on studies of severe sleep deprivation (Fortier-Brochu et al., 2012;
Lim & Dinges, 2010), and there is little to suggest that smaller variations in sleep
behaviours are adversely affecting people. In children, the effects of sleep on
cognition appear to be relatively small (Astill et al., 2012). From a public health point
of view, then, perhaps sleep is not an issue of concern for children in the UK.
Recent studies have shown genetic variation in functional susceptibility to sleep
deprivation in humans (Goel & Dinges, 2012; Pellegrino et al., 2013) and in other
animals (Thimgan, Seugnet, Turk, & Shaw, 2015). These findings fit with the
general theory of differential susceptibility, which posits that a range of genes render
people more or less open to being changed by their environment (Belsky & Pluess,
2009; Ellis, Boyce, Belsky, Bakermans-Kranenburg, & Van IJzendoorn, 2011).
These genes may act through a range of mechanisms including metabolic,
cardiovascular, immune and central nervous system functioning to affect allostasis
(the person’s ability to achieve homeostasis despite challenges to regulation)
(Karatsoreos & McEwen, 2011). It may therefore be that, as well as individual
differences in sleep need, there are individual differences in how well people cope
when their sleep need is not met.
Even in groups of children with disabilities, who are often assumed to be at greater
risk for sleep problems (e.g. Robinson & Richdale, 2004), objective measures may
show that their actual sleep duration is no different from typically developing
children. For example, several studies have replicated the finding that children with
autistic spectrum disorders do not differ from typically developing children in their
total time asleep as measured by actigraphy (Allik, Larsson, & Smedje, 2006;
Souders et al., 2009; Wiggs & Stores, 2004). Ashworth, Hill, Karmiloff-Smith and
Dimitriou (2013) found that children with Down Syndrome and Williams Syndrome
128
slept for the same amount of time per night as a control group, despite parent-
reported problems with bedtime resistance and night-time wakings. One
explanation for the problems reported in this study might be that the children in both
disability groups had an earlier bed time than the control group. Perhaps these
children did not need to go to bed any earlier than their age-matched peers, and if
they had gone to bed at a, “normal,” time they might have experienced fewer
problems.
While there is no doubt that significant sleep deprivation in adults has serious
cognitive consequences, the prevalence of both sleep deprivation and of cognitive
problems resulting from sleep deprivation may be lower than I had believed. The
evidence for children is even less clear. It is hardly surprising, then, that in the
current study a sample of the general population of UK children showed no
significant relationship between sleep and working memory, and no clear
behavioural effects of a universal sleep education intervention.
A distinct contribution
The current research adds to our knowledge about the relationship between sleep
and working memory in children, and also to our understanding of methodological
issues in this area.
Methodology
This study has confirmed the relationship between actigraphic and diary measures
of sleep behaviours. The findings replicate those found in other countries and with
children of higher socioeconomic status (e.g. Tremaine, Dorrian, & Blunden, 2010;
Werner et al., 2008), showing that diary and actigraphy measures are in good
129
agreement about sleep and wake times, but that diary measures consistently
underestimate wake after sleep onset compared to actigraphy.
Problems with using diary measures repeatedly have been found, which have not
been reported elsewhere. Qualitative focus group feedback suggests ways in which
future researchers might improve the return rate for sleep diaries over time, which is
also new to the literature in this field.
A completely novel finding from this study is that wearing an actigraph did not affect
how children completed their sleep diaries. It is, perhaps, surprising that no data
has been published to date relating to this issue. It would be a reasonable lay
prediction that children would be more accurate in reporting their sleep behaviours
when they knew that objective assessment was taking place. However, the current
study shows that children of this age filled in their sleep diaries with similar variation
whether or not they were given an actiwatch to wear.
This study has also provided the field with an adapted sleep knowledge
questionnaire which is appropriate for primary aged children and which has better
psychometric properties than the one it was based upon (Benveniste, Thompson, &
Blunden, 2012). This questionnaire could be used in future research or for
practitioners to evaluate sleep education programmes in their settings.
The relationship between sleep and working memory in children
This study adds correlational data to the body of knowledge about sleep and
working memory in children. Most variables showed no significant correlation with
each other. The only significant results were a negative linear relationship between
diary sleep duration and verbal WM (shorter sleep was related to higher verbal WM
130
scores), and a positive linear relationship between actigraphic sleep efficiency and
visual WM when socioeconomic status was partialled out. The correlational data is
drawn from a large sample of children who were not self-selected, making the
findings relatively robust. These results represent a strong challenge to the
assertion of Kopasz et al. (2010) that, “most studies support the hypothesis that
sleep facilitates working memory as well as memory consolidation in children and
adolescents” (p.167).
The intervention failed to produce systematic effects on sleep variables, so the
experimental data could not show whether changes in sleep were related to
changes in working memory performance.
To explain these findings, a neurodevelopmental perspective may be helpful. The
neural mechanisms underlying working memory are at a relatively early stage of
development in this age group. If sleep has similar effects on the brain throughout
the lifecourse, it is possible that working memory might not be strongly related to
sleep until the fronto-parietal networks subserving it are mature. Another way to
look at these results is to consider the body’s ability for allostasis. Animals including
humans have a range of mechanisms to help compensate for challenges to
homeostasis (Karatsoreos & McEwen, 2011), and therefore small changes in sleep
behaviours may be well tolerated in most people.
The future for sleep interventions
The current sleep intervention was effective in increasing sleep knowledge, an effect
which was maintained at the two to three month follow-up point. However, this
knowledge did not translate into changed sleep behaviour as far as the data in this
study can show. This pattern is a common feature of universal sleep education
131
interventions (Blunden et al., 2012). Indeed, it is not clear what behaviours would
constitute a good outcome for a sleep intervention. Our theoretical understanding of
optimal sleep, at different developmental stages and for different individuals, is not
yet sufficiently developed for us to be able to define behavioural targets. In the
current study, average sleep duration and efficiency were measured, but it may be
that other variables such as daily variations are more important to target.
While there may be longer-term and wider benefits than were measured in this
study, the immediate results suggest that universal sleep education programmes for
this age range are not a good use of resources, largely because most children are
getting enough sleep without any intervention. Instead, sleep interventions could
focus on helping children who are already starting to develop sleep problems (e.g.
Quach et al., 2013) and could use other theoretical underpinnings to support
behavioural change (e.g. motivational interviewing; see Miller & Rose, 2009) or
change in perceptions of problems (e.g. solution focused therapy approaches to
changing, “viewing,” as well as, “doing;” see Bannink, 2007).
Since this study was conducted, a small-scale study using motivational interviewing
with junior-aged children has been published (Willgerodt, Kieckhefer, Ward, & Lentz,
2014). The research was presented as a feasibility study, and the authors found
that using actigraphy to inform motivational interviewing cycles was acceptable to
the families in their study. Willgerodt et al. (2014) were able to recruit nine families
who wished to improve their child’s sleep (the response rate is unknown as the
number of families invited to participate was not reported). These families
participated in four episodes of data-gathering using sleep diaries and actigraphy,
two of which also included motivational interviewing with parents and child together
to set goals based on the data gathered. Three of the children in this study had a
disability (attention deficit hyperactivity disorder or learning disability), and all three
132
had problems engaging in both the data collection and intervention procedures. As
a feasibility study, it shows that a short intervention using this approach is
manageable and acceptable for practitioners and for families with children who do
not have a disability. Unfortunately the results are not reported in enough detail to
see whether the intervention was effective, even for the six families who participated
in the whole protocol.
In terms of targeting children who may benefit from sleep interventions, the
Willgerodt et al. (2014) study used a self-selected group of families with a higher
than average socio-economic status (all the mothers were educated to degree
level). Perhaps a more inclusive way to identify children would be to screen a
population and offer intervention to those experiencing sleep problems of some kind.
Objective methods such as actigraphy would not be feasible due to the cost of
providing large numbers of devices, and using a simple cut-off for short sleep would
include children who need less sleep while excluding those with average sleep but
with other sleep problems such as bedtime resistance. Perhaps the most practical
way to target children for sleep interventions would be a screening questionnaire
such as BEARS (Bedtime issues, Excessive daytime sleepiness, Awakenings during
the night, Regularity and duration of sleep, Snoring; Owens & Dalzell, 2005) or
CSHQ (Children's Sleep Habits Questionnaire; Owens, Spirito, & McGuinn, 2001).
Such a questionnaire could help to identify the kind of help that might be appropriate
for each individual, for example snoring might require medical attention while
regularity of sleep routine would be more amenable to behavioural approaches.
A further issue for sleep interventions is the subjective nature of sleep problems.
What parents report as a sleep problem is likely to be highly culturally bound (Jenni
& O'Connor, 2005), which means that interventions could focus on perceptions of
problems as well as trying to change the sleep behaviours themselves. For
133
example, a child may be a short sleeper but not suffer any functional impairments
such as daytime sleepiness or poor vigilance. Interventions could focus on helping
such a family to accept individual differences in sleep need, worry less about getting
the child to sleep more and find ways to manage the child who wakes early or stays
up late. In addition, the behaviours advocated in sleep intervention programmes
need to be culturally sensitive. For example, good sleep hygiene in the USA might
include a consistent bedtime routine and a low-stimulus sleep environment, but in
other contexts this advice would be at odds with most families who allow their
children to stay up with the family at night (Ottaviano et al., 1996) or who have
several family members sleeping in one room (Oka, Suzuki, & Inoue, 2008).
The future for sleep research
It is difficult ethically to reproduce the sleep research methods commonly used in
adults (e.g. severe sleep deprivation) when working with child participants.
However, the current study suggests some ways in which research might usefully
continue the search for understanding about how sleep affects cognition in children.
Variability within individuals from night to night, and from week to week, was
surprisingly wide in the current sample. The sources of this variation would be
useful to identify. In addition, these natural variations provide an opportunity to
study acute changes in cognitive performance as a function of sleep variables from
the previous night. Individual differences could be a useful approach to future
research, given the possibility that different children will have different levels of sleep
need, which means that there may be no group-level relationship between sleep and
cognitive variables.
134
Another avenue might be to screen participants so that children with short sleep
durations can be selected for targeted study. In adults, Grandner, Patel, Gehrman,
Perlis and Pack (2010) suggest that people who habitually sleep less than six hours
per night would be a helpful group to study, including two groups: those who are
short sleepers because they need less sleep than average and those who are short
sleepers due to chronic sleep deprivation. Defining short sleep in this way may be a
helpful rule of thumb, but it suffers from methodological difficulties (for example, the
six-hour cutoff would be likely to include more people if measured by diary or
questionnaire than by actigraphy), and includes a large proportion of the population
(24.6% in a 1990 survey of over 30,000 US adults; Hale & Do, 2007). In children,
the definition of short sleep would vary by age, but within a study it would be
possible to identify a group. Gradisar et al. (2008) split their participants into three
roughly equal groups according to their sleep duration, but for a targeted study it
may be preferable to focus on a smaller proportion than one-third of the population
in order to see significant differences between short sleepers and other children.
Such an approach might be able to identify group-level correlations between sleep
parameters and working memory measures, as it would seem that cognitive
functions are resilient to smaller variations in sleep duration. In addition, targeting a
group of short sleepers would enable an experimental approach such as a sleep
education programme to have a greater chance of changing sleep behaviours than
the universal approach used in the current study.
In terms of the neurodevelopmental perspective on sleep and working memory,
future research could take a longitudinal approach and include imaging methods in
order to track the development of the frontoparietal networks involved in adult
working memory performance. It would be useful to investigate the extent of sleep
deprivation required before working memory becomes significantly impaired, at
different stages of development. Individual differences in vulnerability to sleep
135
deprivation could also be a fruitful avenue for investigation, including whether any
such differences are stable traits over the course of development. In addition, it
would be helpful to know whether sleep extension can improve working memory
performance or change the neural activity underlying working memory tasks,
especially in children with habitual short sleep durations. One specific
neuropsychological hypothesis to test would be that the relationship between sleep
and WM becomes stronger as the individual’s brain matures into using fronto-
parietal areas that may be more susceptible to the effects of sleep deprivation.
Key messages for educational psychologists
From this literature review and empirical study, a number of key points can be
summarised for practitioner educational psychologists:
There is wide variation in what could be considered normal sleep duration for
children. There are differences between individuals, but also within an
individual from night to night.
Brief night-time wakings are very frequent in children. If parents are
reporting problems in this area, the psychologist needs to consider the
context (e.g. if the parent is in the child’s room noticing every movement,
they may simply be reporting normal wakings).
Children with developmental disabilities are not necessarily more likely to
have sleep problems. There is a complex interplay between family
expectations and child behaviour, for example it may be that a child with a
disability such as Down Syndrome is being asked to go to bed earlier than is
physiologically needed.
Assessment of sleep by diary method is likely to give a reasonably accurate
picture of sleep onset and offset times, but not of night-time wakings. To
take these wakings into account, actigraphy would be a useful method.
136
Since this study was started, much cheaper and more accessible forms of
actigraphy have become commercially available (e.g. FitBit, Jawbone and
Garmin brands). Although their accuracy would need to be established, a
device of this sort may be helpful for a psychologist to have in their library of
equipment.
Most children of primary school age are relatively resilient to sleep
deprivation, although there are likely to be individual differences in this
resilience.
Sleep duration has a weak relationship at best with working memory in
children. Advocating that all children get more sleep is unlikely to produce
significant gains in cognitive skills of this sort. However, there may well be
other benefits of adequate sleep such as reduced risk of obesity and better
emotional regulation, which were not addressed in this study.
Universal interventions may not be an appropriate way to address sleep
difficulties in the primary school age range. From the current study, it seems
that most children in the general population did not need to change their
sleep habits, and for those who did, a short universal education programme
may not be enough to alter their sleep behaviours.
For individual casework, it is worth asking questions about sleep as part of
the psychological formulation. If sleep problems are identified, the
psychologist needs to consider whether the issues are around family
perceptions and expectations, actual sleep behaviours or both. Sleep may
only need to be prioritised for intervention if it appears to be affecting
daytime functioning in some way.
137
Reference List
Acebo, C., Sadeh, A., Seifer, R., Tzischinsky, O., Wolfson, A. R., Hafer, A. et
al. (1999). Estimating sleep patterns with activity monitoring in children and
adolescents: how many nights are necessary for reliable measures? Sleep, 22, 95-
103.
Ackerman, P. L., Beier, M. E., & Boyle, M. O. (2005). Working memory and
intelligence: The same or different constructs? Psychological Bulletin, 131, 30-60.
Ajzen, I. & Fishbein, M. (1980). Understanding attitudes and predicting social
behaviour. New Jersey: Prentice-Hall.
Allik, H., Larsson, J. O., & Smedje, H. (2006). Sleep patterns of school-age
children with Asperger syndrome or high-functioning autism. Journal of Autism and
Developmental Disorders, 36, 585-595.
Alloway, T. (2012). Can interactive working memory training improve
learning? Journal of Interactive Learning Research, 23, 197-207.
Alloway, T. P. (2007). Automated Working Memory Assessment. Oxford:
Pearson Assessment.
Alloway, T. P. & Gathercole, S. E. (2006). How does working memory work
in the classroom? Educational Research & Reviews, 1, 134-139.
Alloway, T. P., Gathercole, S. E., & Pickering, S. J. (2006). Verbal and visuo-
spatial short-term and working memory in children: Are they separable? Child
Development, 77, 1698-1716.
Alloway, T. P. (2009). Working memory, but not IQ, predicts subsequent
learning in children with learning difficulties. European Journal of Psychological
138
Assessment, 25, 92.
Alloway, T. P. & Alloway, R. G. (2010). Investigating the predictive roles of
working memory and IQ in academic attainment. Journal of Experimental Child
Psychology, 106, 20-29.
Armitage, C. J. & Conner, M. (2001). Efficacy of the theory of planned
behaviour: A meta-analytic review. British Journal of Social Psychology, 40, 471-
499.
Ashworth, A., Hill, C. M., Karmiloff-Smith, A., & Dimitriou, D. (2013). Cross
syndrome comparison of sleep problems in children with Down syndrome and
Williams syndrome. Research in Developmental Disabilities, 34, 1572-1580.
Astill, R. G., Van der Heijden, K. B., Van IJzendoorn, M. H., & Van Someren,
E. J. (2012). Sleep, cognition, and behavioral problems in school-age children: A
century of research meta-analyzed. Psychological Bulletin, 138, 1109-1137.
Baddeley, A. (2012). Working memory: Theories, models, and controversies.
Annual Review of Psychology, 63, 1-29.
Baddeley, A. (2000). The episodic buffer: a new component of working
memory? Trends in Cognitive Sciences, 4, 417-423.
Bannink, F. P. (2007). Solution-focused brief therapy. Journal of
Contemporary Psychotherapy, 37, 87-94.
Barbey, A. K., Koenigs, M., & Grafman, J. (2013). Dorsolateral prefrontal
contributions to human working memory. Cortex, 49, 1195-1205.
Beebe, D. W., Difrancesco, M. W., Tlustos, S. J., McNally, K. A., & Holland,
S. K. (2009). Preliminary fMRI findings in experimentally sleep-restricted
139
adolescents engaged in a working memory task. Behavioural and Brain Functions,
5, 9.
Beebe, D. W. (2012). A brief primer on sleep for pediatric and child clinical
neuropsychologists. Child Neuropsychology, 18, 313-338.
Belsky, J. & Pluess, M. (2009). Beyond diathesis stress: differential
susceptibility to environmental influences. Psychological Bulletin, 135, 885-908.
Benveniste, T. C., Thompson, K., & Blunden, S. L. (2012). With age comes
knowledge? Sleep knowledge in Australian children. In X. Zhou & C. Sargent (Eds.),
Sleep of Different Populations (pp. 11-15). Adelaide: Australasian Chronobiology
Society.
Bin, Y. S., Marshall, N. S., & Glozier, N. (2012). Secular trends in adult sleep
duration: a systematic review. Sleep Medicine Reviews, 16, 223-230.
Blair, C. & Razza, R. P. (2007). Relating effortful control, executive function,
and false belief understanding to emerging math and literacy ability in kindergarten.
Child Development, 78, 647-663.
Blasi, G., Selvaggi, P., Fazio, L., Antonucci, L. A., Taurisano, P., Masellis, R.
et al. (2015). Variation in dopamine D2 and serotonin 5-HT2A receptor genes is
associated with working memory processing and response to treatment with
antipsychotics. Neuropsychopharmacology, 40, 1600-1608.
Blunden, S. (2007a). The implementation of a sleep education program in
adolescents. Sleep and Biological Rhythms, 5, A31.
Blunden, S. (2007b). The implementation of a sleep education program in
primary school children. Sleep and Biological Rhythms, 5, A32.
140
Blunden, S., Kira, G., Hull, M., & Maddison, R. (2013). Does sleep education
change sleep parameters? Comparing sleep education trials for middle school
students in Australia and New Zealand. The Open Sleep Journal, 5, 12-18.
Blunden, S. L., Chapman, J., & Rigney, G. A. (2012). Are sleep education
programs successful? The case for improved and consistent research efforts. Sleep
Medicine Reviews, 16, 355-370.
Braun, A. R., Balkin, T. J., Wesenten, N. J., Carson, R. E., Varga, M.,
Baldwin, P. et al. (1997). Regional cerebral blood flow throughout the sleep-wake
cycle. An H2 (15) O PET study. Brain, 120, 1173-1197.
Brewer, M. B. (2000). Research design and issues of validity. In H.Reis & C.
Judd (Eds.), Handbook of Research Methods in Social and Personality Psychology
(pp. 3-16). Cambridge: Cambridge University Press.
Brickenkamp, R. & Zilmer, E. (1998). d2 Test of Attention (English language
version). Gottingen: Hogrefe Publishing.
Brunoni, A. R. & Vanderhasselt, M. A. (2014). Working memory improvement
with non-invasive brain stimulation of the dorsolateral prefrontal cortex: A systematic
review and meta-analysis. Brain and Cognition, 86, 1-9.
Buckhalt, J. A., El-Sheikh, M., & Keller, P. (2007). Children's sleep and
cognitive functioning: race and socioeconomic status as moderators of effects. Child
Development, 78, 213-231.
Bull, R., Espy, K. A., & Wiebe, S. A. (2008). Short-term memory, working
memory, and executive functioning in preschoolers: Longitudinal predictors of
mathematical achievement at age 7 years. Developmental Neuropsychology, 33,
205-228.
141
Bull, R. & Scerif, G. (2001). Executive functioning as a predictor of children's
mathematics ability: Inhibition, switching, and working memory. Developmental
Neuropsychology, 19, 273-293.
Cain, K., Oakhill, J., & Bryant, P. (2004). Children's reading comprehension
ability: Concurrent prediction by working memory, verbal ability and component
skills. Journal of Educational Psychology, 96, 31-42.
Calhoun, S. L., Fernandez-Mendoza, J., Vgontzas, A. N., Mayes, S. D.,
Tsaoussoglou, M., Rodriguez-Munoz, A. et al. (2012). Learning,
attention/hyperactivity, and conduct problems as sequelae of excessive daytime
sleepiness in a general population study of young children. Sleep, 35, 627-632.
Carskadon, M. A. & Dement, W. C. (2011). Monitoring and staging human
sleep. In M.H.Kryger, T. Roth, & W. C. Dement (Eds.), Principles and practice of
sleep medicine (5th edition) (pp. 16-26). St Louis: Elsevier Saunders.
Casey, B. J., Tottenham, N., Liston, C., & Durston, S. (2005). Imaging the
developing brain: what have we learned about cognitive development? Trends in
Cognitive Sciences, 9, 104-110.
Chaddock, L., Erickson, K. I., Prakash, R. S., Kim, J. S., Voss, M. W.,
VanPatter, M. et al. (2010). A neuroimaging investigation of the association between
aerobic fitness, hippocampal volume, and memory performance in preadolescent
children. Brain Research, 1358, 172-183.
Chee, M. W. L., Chuah, L. Y. M., Venkatraman, V., Chan, W. Y., Philip, P., &
Dinges, D. F. (2006). Functional imaging of working memory following normal sleep
and after 24 and 35 h of sleep deprivation: Correlations of fronto-parietal activation
with performance. Neuroimage, 31, 419-428.
142
Chee, M. W. & Choo, W. C. (2004). Functional imaging of working memory
after 24 hr of total sleep deprivation. The Journal of Neuroscience, 24, 4560-4567.
Chen, X., Beydoun, M. A., & Wang, Y. (2008). Is sleep duration associated
with childhood obesity? A systematic review and meta-analysis. Obesity, 16, 265-
274.
Cho, M., Quach, J., Anderson, P., Mensah, F., Wake, M., & Roberts, G.
(2015). Poor Sleep and Lower Working Memory in Grade 1 Children: Cross-
Sectional, Population-Based Study. Academic Pediatrics, 15, 111-116.
Choo, W. C., Lee, W. W., Venkatraman, V., Sheu, F. S., & Chee, M. W.
(2005). Dissociation of cortical regions modulated by both working memory load and
sleep deprivation and by sleep deprivation alone. Neuroimage, 25, 579-587.
Chun, M. M. (2011). Visual working memory as visual attention sustained
internally over time. Neuropsychologia, 49, 1407-1409.
Cirelli, C. & Tononi, G. (2008). Is sleep essential? PLoS Biology, 6, 1605-
1611.
Cohen, J. (1992). A power primer. Psychological Bulletin, 112, 155-159.
Colrain, I. M. & Baker, F. C. (2011). Changes in sleep as a function of
adolescent development. Neuropsychology Review, 21, 5-21.
Conway, A. R. A., Cowan, N., Bunting, M. F., Therriault, D. J., & Minkoff, S.
R. B. (2002). A latent variable analysis of working memory capacity, short-term
memory capacity, processing speed and general fluid intelligence. Intelligence, 30,
163-183.
Conway, A. R. A., Kane, M. J., & Engle, R. W. (2003). Working memory
143
capacity and its relation to general intelligence. Trends in Cognitive Sciences, 7,
547-552.
Cowan, N. & Chen, Z. (2009). How chunks form in long-term memory and
affect short-term memory limits. In A.Thorn & M. Page (Eds.), Interactions between
short-term and long-term memory in the verbal domain (pp. 86-101). Hove:
Psychology Press.
Crabtree, V. & Williams, N. (2009). Normal sleep in children & adolescents.
Child and Adolescent Psychiatric Clinics of North America, 18, 799-811.
Crotty, M. (1998). The foundations of social research: Meaning and
perspective in the research process. London: Sage.
Dadds, M. R. & Rhodes, T. (2008). Aggression in young children with
concurrent callous-unemotional traits: can the neurosciences inform progress and
innovation in treatment approaches? Philosophical Transactions of the Royal
Society B: Biological Sciences, 363, 2567-2576.
Damery, S., Ryan, R., McManus, R. J., Warmington, S., Draper, H., &
Wilson, S. (2011). The effect of seeking consent on the representativeness of
patient cohorts: iron-deficiency anaemia and colorectal cancer. Colorectal Disease,
13, e366-e373.
Darki, F. & Klingberg, T. (2015). The role of fronto-parietal and fronto-striatal
networks in the development of working memory: a longitudinal study. Cerebral
Cortex, 25, 1587-1595.
De Sonneville, L. M. J. (1999). Amsterdam Neuropsychological Tasks: A
computer-aided assessment program. Computers in Psychology, 6, 187-203.
Department for Communities & Local Government (2011). The English
144
Indices of Deprivation 2010 Downloaded 20th March 2015 from
https://www.gov.uk/government/statistics/english-indices-of-deprivation-2010.
Dewald, J. F., Meijer, A. M., Oort, F. J., Kerkhof, G. A., & Bogels, S. M.
(2010). The influence of sleep quality, sleep duration and sleepiness on school
performance in children and adolescents: A meta-analytic review. Sleep Medicine
Reviews, 14, 179-189.
Diamond, A. & Lee, K. (2011). Interventions shown to aid executive function
development in children 4 to 12 years old. Science, 333, 959-964.
Diekelmann, S. & Born, J. (2010). The memory function of sleep. Nature
Reviews Neuroscience, 11, 114-126.
Drummond, S. P. A. & Brown, G. G. (2001). The effects of total sleep
deprivation on cerebral responses to cognitive performance.
Neuropsychopharmacology, 25, S68-S73.
Dunning, D. L. & Holmes, J. (2014). Does working memory training promote
the use of strategies on untrained working memory tasks? Memory & Cognition, 1-9.
Durmer, J. S. & Dinges, D. F. (2005). Neurocognitive consequences of sleep
deprivation. Seminars in Neurology, 25, 117-129.
Dwan, K., Altman, D. G., Arnaiz, J. A., Bloom, J., Chan, A. W., Cronin, E. et
al. (2008). Systematic review of the empirical evidence of study publication bias and
outcome reporting bias. PLoS One, 3, e3081.
Ellis, B. J., Boyce, W. T., Belsky, J., Bakermans-Kranenburg, M. J., & Van
IJzendoorn, M. H. (2011). Differential susceptibility to the environment: An
evolutionary-neurodevelopmental theory. Development and Psychopathology, 23, 7-
28.
145
Engel, P. M. J., Santos, F. H. s., & Gathercole, S. E. (2008). Are working
memory measures free of socioeconomic influence? Journal of Speech, Language,
and Hearing Research, 51, 1580-1587.
Ferrie, J. E., Shipley, M. J., Akbaraly, T. N., Marmot, M. G., Kivimaki, M., &
Singh-Manoux, A. (2011). Change in Sleep Duration and Cognitive Function:
Findings from the Whitehall II Study. Sleep, 34, 565-573.
Field, A. (2013). Discovering statistics using IBM SPSS Statistics, 4th ed.
London: Sage.
Finn, A. S., Sheridan, M. A., Kam, C. L. H., Hinshaw, S., & D'Esposito, M.
(2010). Longitudinal evidence for functional specialization of the neural circuit
supporting working memory in the human brain. The Journal of Neuroscience, 30,
11062-11067.
Fortier-Brochu, E., Beaulieu-Bonneau, S., Ivers, H., & Morin, C. M. (2012).
Insomnia and daytime cognitive performance: A meta-analysis. Sleep Medicine
Reviews, 16, 83-94.
Galland, B. C., Taylor, B. J., Elder, D. E., & Herbison, P. (2012). Normal
sleep patterns in infants and children: A systematic review of observational studies.
Sleep Medicine Reviews, 16, 213-222.
Gathercole, S. E., Dunning, D. L., & Holmes, J. (2012). Cogmed training:
Let's be realistic about intervention research. Journal of Applied Research in
Memory and Cognition, 1, 201-203.
Gathercole, S. E., Pickering, S. J., Knight, C., & Stegmann, Z. (2004).
Working memory skills and educational attainment: evidence from national
curriculum assessments at 7 and 14 years of age. Applied Cognitive Psychology,
146
18, 1-16.
Gathercole, S. E., Alloway, T. P., Willis, C., & Adams, A. M. (2006). Working
memory in children with reading disabilities. Journal of Experimental Child
Psychology, 93, 265-281.
Geiger, A., Achermann, P., & Jenni, O. G. (2010). Association between sleep
duration and intelligence scores in healthy children. Developmental Psychology, 46,
949.
Geiger-Brown, J., Trinkoff, A., & Rogers, V. E. (2011). The impact of work
schedules, home, and work demands on self-reported sleep in registered nurses.
Journal of Occupational and Environmental Medicine, 53, 303-307.
Gelao, B., Fazio, L., Selvaggi, P., Di Giorgio, A., Taurisano, P., Quarto, T. et
al. (2014). DRD2 genotype predicts prefrontal activity during working memory after
stimulation of D2 receptors with bromocriptine. Psychopharmacology, 231, 2361-
2370.
Goel, N. & Dinges, D. F. (2012). Predicting risk in space: genetic markers for
differential vulnerability to sleep restriction. Acta Astronautica, 77, 207-213.
Goodwin, L. & Leech, N. (2006). Understanding correlation: Factors that
affect the size of r. The Journal of Experimental Education, 74, 251-266.
Gough, D. (2007). Weight of evidence: a framework for the appraisal of the
quality and relevance of evidence. Research Papers in Education, 22, 213-228.
Gozal, D. & Kheirandish-Gozal, L. (2012). Childhood obesity and sleep:
relatives, partners, or both? A critical perspective on the evidence. Annals of the
New York Academy of Sciences, 1264, 135-141.
147
Gradisar, M., Gardner, G., & Dohnt, H. (2011). Recent worldwide sleep
patterns and problems during adolescence: A review and meta-analysis of age,
region and sleep. Sleep Medicine, 12, 110-118.
Gradisar, M., Terrill, G., Johnston, A., & Douglas, P. (2008). Adolescent
sleep and working memory performance. Sleep and Biological Rhythms, 6, 146-154.
Grandner, M. A., Patel, N. P., Gehrman, P. R., Perlis, M. L., & Pack, A. I.
(2010). Problems associated with short sleep: bridging the gap between laboratory
and epidemiological studies. Sleep Medicine Reviews, 14, 239-247.
Groff, R. (2004). Critical realism, post-positivism and the possibility of
knowledge. London: Routledge.
Gruber, R., Cassoff, J., Frenette, S., Wiebe, S., & Carrier, J. (2012). Impact
of sleep extension and restriction on children's emotional lability and impulsivity.
Pediatrics, 130, e1155-e1161.
Gruber, R., Laviolette, R., Deluca, P., Monson, E., Cornish, K., & Carrier, J.
(2010). Short sleep duration is associated with poor performance on IQ measures in
healthy school-age children. Sleep Medicine, 11, 289-294.
Hagewoud, R., Havekes, R., Novati, A., Keijser, J., Van Der Zee, E., &
Meerlo, P. (2010). Sleep deprivation impairs spatial working memory and reduces
hippocampal AMPA receptor phosphorylation. Journal of Sleep Research, 19, 280-
288.
Haier, R. J., Siegel, B. V., Nuechterlein, K. H., Hazlett, E., Wu, J. C., Paek, J.
et al. (1988). Cortical glucose metabolic rate correlates of abstract reasoning and
attention studied with positron emission tomography. Intelligence, 12, 199-217.
Hairston, I. S., Little, M. T. M., Scanlon, M. D., Barakat, M. T., Palmer, T. D.,
148
Sapolsky, R. M. et al. (2005). Sleep Restriction Suppresses Neurogenesis Induced
by Hippocampus-Dependent Learning. Journal of Neurophysiology, 94, 4224-4233.
Hale, L. & Do, D. P. (2007). Racial differences in self-reports of sleep
duration in a population-based study. Sleep, 30, 1096-1103.
Hale, L. & Guan, S. (2015). Screen time and sleep among school-aged
children and adolescents: A systematic literature review. Sleep Medicine Reviews,
21, 50-58.
Hardeman, W., Johnston, M., Johnston, D., Bonetti, D., Wareham, N., &
Kinmonth, A. L. (2002). Application of the theory of planned behaviour in behaviour
change interventions: A systematic review. Psychology and Health, 17, 123-158.
Henderson, L. M., Weighall, A. R., Brown, H., & Gareth Gaskell, M. (2012).
Consolidation of vocabulary is associated with sleep in children. Developmental
Science, 15, 674-687.
Holley, S., Hill, C. M., & Stevenson, J. (2009). A comparison of actigraphy
and parental report of sleep habits in typically developing children aged 6 to 11
Years. Behavioral Sleep Medicine, 8, 16-27.
Horovitz, S. G., Braun, A. R., Carr, W. S., Picchioni, D., Balkin, T. J.,
Fukunaga, M. et al. (2009). Decoupling of the brain's default mode network during
deep sleep. Proceedings of the National Academy of Sciences, 106, 11376-11381.
Howard-Jones, P. (2014). Neuroscience and education: A review of
educational interventions and approaches informed by neuroscience. London:
Education Endowment Foundation.
Hulme, C. & Melby-Lervag, M. (2012). Current evidence does not support
the claims made for CogMed working memory training. Journal of Applied Research
149
in Memory and Cognition, 1, 197-200.
Iglowstein, I., Jenni, O. G., Molinari, L., & Largo, R. H. (2003). Sleep duration
from infancy to adolescence: Reference values and generational trends. Pediatrics,
111, 302-307.
Ioannidis, J. P., Munafo, M. R., Fusar-Poli, P., Nosek, B. A., & David, S. P.
(2014). Publication and other reporting biases in cognitive sciences: detection,
prevalence, and prevention. Trends in Cognitive Sciences, 18, 235-241.
Jeneson, A. & Squire, L. (2012). Working memory, long-term memory, and
medial temporal lobe function. Learning and Memory, 19, 15-25.
Jenni, O. G. & O'Connor, B. B. (2005). Children's sleep: an interplay
between culture and biology. Pediatrics, 115, 204-216.
Jiang, F., VanDyke, R. D., Zhang, J., Li, F., Gozal, D., & Shen, X. (2011).
Effect of chronic sleep restriction on sleepiness and working memory in adolescents
and young adults. Journal of Clinical and Experimental Neuropsychology, 33, 892-
900.
Kahn, A., Van de Merckt, C., Rebuffat, E., Mozin, M. J., Sottiaux, M., Blum,
D. et al. (1989). Sleep problems in healthy preadolescents. Pediatrics, 84, 542-546.
Karatsoreos, I. N. & McEwen, B. S. (2011). Psychobiological allostasis:
resistance, resilience and vulnerability. Trends in Cognitive Sciences, 15, 576-584.
Killgore, W. D. S. (2010). Effects of sleep deprivation on cognition. Progress
in Brain Research, 185, 105-129.
Kim, E., Grover, L. M., Bertolotti, D., & Green, T. L. (2010). Growth hormone
rescues hippocampal synaptic function after sleep deprivation. American Journal of
150
Physiology - Regulatory, Integrative and Comparative Physiology, 298, R1588-
R1596.
Kira, G., Maddison, R., Hull, M., Blunden, S., & Olds, T. (2013). Sleep
education improves the sleep duration of adolescents: A randomized controlled pilot
study. Journal of Clinical Sleep Medicine, 10, 787-792.
Kirby, M., Maggi, S., & D'Angiulli, A. (2011). School Start Times and the
Sleep-Wake Cycle of Adolescents: A Review and Critical Evaluation of Available
Evidence. Educational Researcher, 40, 56-61.
Klingberg, T., Forssberg, H., & Westerberg, H. (2002). Training of working
memory in children with ADHD. Journal of Clinical and Experimental
Neuropsychology, 24, 781-791.
Konen, T., Dirk, J., & Schmiedek, F. (2015). Cognitive benefits of last night's
sleep: daily variations in children's sleep behavior are related to working memory
fluctuations. Journal of Child Psychology and Psychiatry, 56, 171-182.
Kopasz, M., Loessl, B., Hornyak, M., Riemann, D., Nissen, C., Piosczyk, H.
et al. (2010). Sleep and memory in healthy children and adolescents - a critical
review. Sleep Medicine Reviews, 14, 167-177.
Laberge, L., Petit, D., Simard, C., Vitaro, F., Tremblay, R. E., & Montplaisir,
J. (2001). Development of sleep patterns in early adolescence. Journal of Sleep
Research, 10, 59-67.
LeBourgeois, M. K., Giannotti, F., Cortesi, F., Wolfson, A. R., & Harsh, J.
(2005). The relationship between reported sleep quality and sleep hygiene in Italian
and American adolescents. Pediatrics, 115, 257-265.
Lee, J. & Park, S. (2005). Working memory impairments in schizophrenia: a
151
meta-analysis. Journal of Abnormal Psychology, 114, 599.
Lenroot, R. K. & Giedd, J. N. (2010). The structural development of the
human brain as measured longitudinally with magnetic resonance imaging. In
D.Coch, K. W. Fischer, & G. Dawson (Eds.), Human Behavior, Learning, and the
Developing Brain: Typical Development (pp. 50-73). New York: Guilford Press.
Levin, H. S. & Hanten, G. (2005). Executive functions after traumatic brain
injury in children. Pediatric Neurology, 33, 79-93.
Levy, R. & Goldman-Rakic, P. S. (1999). Association of storage and
processing functions in the dorsolateral prefrontal cortex of the nonhuman primate.
The Journal of Neuroscience, 19, 5149-5158.
Li, S., Jin, X., Yan, C., Wu, S., Jiang, F., & Shen, X. (2008). Bed-and room-
sharing in Chinese school-aged children: Prevalence and association with sleep
behaviors. Sleep Medicine, 9, 555-563.
Lim, J. & Dinges, D. F. (2008). Sleep deprivation and vigilant attention.
Annals of the New York Academy of Sciences, 1129, 305-322.
Lim, J. & Dinges, D. F. (2010). A meta-analysis of the impact of short-term
sleep deprivation on cognitive variables. Psychological Bulletin, 136, 375-389.
Liu, X., Liu, L., Owens, J. A., & Kaplan, D. L. (2005). Sleep patterns and
sleep problems among schoolchildren in the United States and China. Pediatrics,
115, 241-249.
Liu, X., Liu, L., & Wang, R. (2003). Bed sharing, sleep habits, and sleep
problems among Chinese school-aged children. Sleep, 26, 839-844.
Luber, B., Stanford, A. D., Bulow, P., Nguyen, T., Rakitin, B. C., Habeck, C.
152
et al. (2008). Remediation of sleep-deprivation induced working memory impairment
with fMRI-guided transcranial magnetic stimulation. Cerebral Cortex, 18, 2077-2085.
Luciana, M., Conklin, H. M., Hooper, C. J., & Yarger, R. S. (2005). The
development of nonverbal working memory and executive control processes in
adolescents. Child Development, 76, 697-712.
Lythe, K. E., Williams, S. C., Anderson, C., Libri, V., & Mehta, M. A. (2012).
Frontal and parietal activity after sleep deprivation is dependent on task difficulty
and can be predicted by the fMRI response after normal sleep. Behavioural Brain
Research, 233, 62-70.
Manly, T., Robertson, I., Anderson, V., & Nimmo-Smith, I. (1998). Test of
Everyday Attention for Children. Oxford: Pearson Assessment.
Maquet, P. (2000). Functional neuroimaging of normal human sleep by
positron emission tomography. Journal of Sleep Research, 9, 207-232.
Martinussen, R., Hayden, J., Hogg-Johnson, S., & Tannock, R. (2005). A
meta-analysis of working memory impairments in children with attention-
deficit/hyperactivity disorder. Journal of the American Academy of Child &
Adolescent Psychiatry, 44, 377-384.
Mastin, D. F., Bryson, J., & Corwyn, R. (2006). Assessment of sleep hygiene
using the Sleep Hygiene Index. Journal of behavioral medicine, 29, 223-227.
Matricciani, L., Olds, T., & Petkov, J. (2012). In search of lost sleep: secular
trends in the sleep time of school-aged children and adolescents. Sleep Medicine
Reviews, 16, 203-211.
Matricciani, L. A., Olds, T. S., Blunden, S., Rigney, G., & Williams, M. T.
(2012). Never enough sleep: a brief history of sleep recommendations for children.
153
Pediatrics, 129, 548-556.
McAuley, T. & White, D. e. A. (2011). A latent variables examination of
processing speed, response inhibition, and working memory during typical
development. Journal of Experimental Child Psychology, 108, 453-468.
Melby-Lervag, M. & Hulme, C. (2013). Is working memory training effective?
A meta-analytic review. Developmental Psychology, 49, 270-291.
Meltzer, L. J. & Mindell, J. A. (2007). Relationship between child sleep
disturbances and maternal sleep, mood, and parenting stress: a pilot study. Journal
of Family Psychology, 21, 67-73.
Meltzer, L. J. & Montgomery-Downs, H. E. (2011). Sleep in the family.
Pediatric Clinics of North America, 58, 765-774.
Meltzer, L. J., Montgomery-Downs, H. E., Insana, S. P., & Walsh, C. M.
(2012). Use of actigraphy for assessment in pediatric sleep research. Sleep
Medicine Reviews, 16, 463-475.
Michie, S., Johnston, M., Francis, J., Hardeman, W., & Eccles, M. (2008).
From theory to intervention: mapping theoretically derived behavioural determinants
to behaviour change techniques. Applied Psychology, 57, 660-680.
Miller, W. R. & Rose, G. S. (2009). Toward a theory of motivational
interviewing. American Psychologist, 64, 527-537.
Morrison, A. B. & Chein, J. M. (2011). Does working memory training work?
The promise and challenges of enhancing cognition by training working memory.
Psychonomic Bulletin & Review, 18, 46-60.
Moseley, L. & Gradisar, M. (2009). Evaluation of a school-based intervention
154
for adolescent sleep problems. Sleep, 32, 334-341.
Mu, Q., Nahas, Z., Johnson, K. A., Yamanaka, K., Mishory, A., Koola, J. et
al. (2005). Decreased cortical response to verbal working memory following sleep
deprivation. Sleep, 28, 55-67.
Muzur, A., Pace-Schott, E. F., & Hobson, J. A. (2002). The prefrontal cortex
in sleep. Trends in Cognitive Sciences, 6, 475-481.
Nee, D. E. & Jonides, J. (2011). Dissociable contributions of prefrontal cortex
and the hippocampus to short-term memory: Evidence for a 3-state model of
memory. Neuroimage, 54, 1540-1548.
Oka, Y., Suzuki, S., & Inoue, Y. (2008). Bedtime activities, sleep
environment, and sleep/wake patterns of Japanese elementary school children.
Behavioral Sleep Medicine, 6, 220-233.
Ottaviano, S., Giannotti, F., Cortesi, F., Bruni, O., & Ottaviano, C. (1996).
Sleep characteristics in healthy children from birth to 6 years of age in the urban
area of Rome. Sleep, 19, 1-3.
Owen, A. M., Downes, J. J., Sahakian, B. J., Polkey, C. E., & Robbins, T. W.
(1990). Planning and spatial working memory following frontal lobe lesions in man.
Neuropsychologia, 28, 1021-1034.
Owens, J. A., Belon, K., & Moss, P. (2010). Impact of delaying school start
time on adolescent sleep, mood, and behavior. Archives of Pediatrics & Adolescent
Medicine, 164, 608-614.
Owens, J. A. & Dalzell, V. (2005). Use of the "BEARS" sleep screening tool
in a pediatric residents' continuity clinic: a pilot study. Sleep Medicine, 6, 63-69.
155
Owens, J. A., Spirito, A., & McGuinn, M. (2001). The Children's Sleep Habits
Questionnaire (CSHQ): psychometric properties of a survey instrument for school-
aged children. Sleep, 23, 1043-1051.
Park, S. (2013). Always on and always with mobile tablet devices: A
qualitative study on how young adults negotiate with continuous connected
presence. Bulletin of Science, Technology & Society, 33, 182-190.
Pellegrino, R., Kavakli, I. H., Goel, N., Cardinale, C. J., Dinges, D. F., Kuna,
S. T. et al. (2013). A novel BHLHE41 variant is associated with short sleep and
resistance to sleep deprivation in humans. Sleep, 37, 1327-1336.
Pesonen, A. K., Raikkonen, K., Paavonen, E. J., Heinonen, K., Komsi, N.,
Lahti, J. et al. (2010). Sleep duration and regularity are associated with behavioral
problems in 8-year-old children. International Journal of Behavioral Medicine, 17,
298-305.
Petrides, M. (2000). The role of the mid-dorsolateral prefrontal cortex in
working memory. In W.Schneider, A. Owen, & J. Duncan (Eds.), Executive Control
and the Frontal Lobe: Current Issues (pp. 44-54). Springer Berlin Heidelberg.
Pilcher, J. J. & Huffcutt, A. J. (1996). Effects of sleep deprivation on
performance: a meta-analysis. Sleep: Journal of Sleep Research & Sleep Medicine,
19, 318-326.
Quach, J., Gold, L., Arnup, S., Sia, K. L., Wake, M., & Hiscock, H. (2013).
Sleep well, be well study: improving school transition by improving child sleep: a
translational randomised trial. BMJ Open, 3, e004009.
Quach, J., Hiscock, H., Ukoumunne, O. C., & Wake, M. (2011). A brief sleep
intervention improves outcomes in the school entry year: a randomized controlled
156
trial. Pediatrics, 128, 692-701.
Redick, T. S., Shipstead, Z., Harrison, T. L., Hicks, K. L., Fried, D. E.,
Hambrick, D. Z. et al. (2013). No evidence of intelligence improvement after working
memory training: a randomized, placebo-controlled study. Journal of Experimental
Psychology: General, 142, 359-379.
Robinson, A. M. & Richdale, A. L. (2004). Sleep problems in children with an
intellectual disability: parental perceptions of sleep problems, and views of treatment
effectiveness. Child: Care, Health and Development, 30, 139-150.
Robson, C. (2011). Real world research: a resource for users of social
research methods in applied settings. Chichester: Wiley.
Rottschy, C., Langner, R., Dogan, I., Reetz, K., Laird, A. R., Schulz, J. B. et
al. (2012). Modelling neural correlates of working memory: A coordinate-based
meta-analysis. Neuroimage, 60, 830-846.
Ruben, G.-M., Suntsova, N., Stewart, D. R., Gong, H., Szymusiak, R., &
McGinty, D. (2003). Sleep deprivation reduces proliferation of cells in the dentate
gyrus of the hippocampus in rats. The Journal of Physiology, 549, 563-571.
Ruskin, D. N., Liu, C., Dunn, K. E., Bazan, N. G., & LaHoste, G. J. (2004).
Sleep deprivation impairs hippocampus-mediated contextual learning but not
amygdala-mediated cued learning in rats. European Journal of Neuroscience, 19,
3121-3124.
Sadeh, A., Ravi, A., & Gruber, R. (2000). Sleep patterns and sleep
disruptions in school-aged children. Developmental Psychology, 36, 291-301.
Sadeh, A., Dahl, R. E., Shahar, G., & Rosenblat-Stein, S. (2009). Sleep and
the transition to adolescence: a longitudinal study. Sleep, 32, 1602-1609.
157
Sadeh, A., Gruber, R., & Raviv, A. (2002). Sleep, neurobehavioral
functioning, and behavior problems in school-age children. Child Development, 73,
405-417.
Sadeh, A., Gruber, R., & Raviv, A. (2003). The effects of sleep restriction
and extension on school-age children: What a difference an hour makes. Child
Development, 74, 444-455.
Schwichtenberg, A. J. & Goodlin-Jones, B. (2010). Causes and correlates of
frequent night awakenings in early childhood. International Review of Neurobiology,
93, -177.
Scott, A., Shaw, M., & Joughin, C. (2001). Finding the evidence: a gateway
to the literature in child & adolescent mental health (2nd edition). London: Gaskell
(Royal College of Psychiatrists).
Shipstead, Z., Redick, T. S., & Engle, R. W. (2012). Is working memory
training effective? Psychological Bulletin, 138, 628-654.
Simola, P., Liukkonen, K., Pitkaranta, A., Pirinen, T., & Aronen, E. T. (2014).
Psychosocial and somatic outcomes of sleep problems in children: a 4-year follow-
up study. Child: Care, Health and Development, 40, 60-67.
Smith, G. & Morris, P. (2015). Building confidence in confidence intervals.
The Psychologist, 28, 476-479.
Smith, S. W., Daunic, A. P., & Taylor, G. G. (2007). Treatment fidelity in
applied educational research: Expanding the adoption and application of measures
to ensure evidence-based practice. Education and Treatment of Children, 30, 121-
134.
Souders, M. C., Mason, T. B., Valladares, O., Bucan, M., Levy, S. E.,
158
Mandell, D. S. et al. (2009). Sleep behaviors and sleep quality in children with
autism spectrum disorders. Sleep, 32, 1566-1578.
Spruyt, K., O'Brien, L., Cluydts, R., Verleye, G. B., & Ferri, R. (2005). Odds,
prevalence and predictors of sleep problems in school-age normal children. Journal
of Sleep Research, 14, 163-176.
St Clair-Thompson, H., Stevens, R., Hunt, A., & Bolder, E. (2010). Improving
children's working memory and classroom performance. Educational Psychology,
30, 203-219.
Steenari, M. R., Vuontela, V., Paavonen, E. J., Carlson, S., Fjallberg, M., &
Aronen, E. (2003). Working memory and sleep in 6- to 13-year-old schoolchildren.
Journal of the American Academy of Child & Adolescent Psychiatry, 42, 85-92.
Stein, M. A., Mendelsohn, J., Obermeyer, W. H., Amromin, J., & Benca, R.
(2001). Sleep and behavior problems in school-aged children. Pediatrics, 107, e60.
Such, E. & Walker, R. (2004). Being responsible and responsible beings:
Children's understanding of responsibility. Children and Society, 18, 231-242.
Suss, H.-M., Oberauer, K., Wittmann, W. W., Wilhelm, O., & Schulze, R.
(2002). Working-memory capacity explains reasoning ability - and a little bit more.
Intelligence, 30, 261-288.
Szymczak, J. T., Jasinska, M., Pawlak, E., & Zwierzykowska, M. (1993).
Annual and weekly changes in the sleep-wake rhythm of school children. Sleep, 16,
433-435.
Taki, Y., Hashizume, H., Thyreau, B., Sassa, Y., Takeuchi, H., Wu, K. et al.
(2012). Sleep duration during weekdays affects hippocampal gray matter volume in
healthy children. Neuroimage, 60, 471-475.
159
Tamnes, C. K., Walhovd, K. B., Grydeland, H., Holland, D., Ostby, Y., Dale,
A. M. et al. (2013). Longitudinal working memory development is related to structural
maturation of frontal and parietal cortices. Journal of Cognitive Neuroscience, 25,
1611-1623.
Tassi, P. & Muzet, A. (2001). Defining the states of consciousness.
Neuroscience & Biobehavioral Reviews, 25, 175-191.
Tavernier, R. & Willoughby, T. (2014). Sleep problems: predictor or outcome
of media use among emerging adults at university? Journal of Sleep Research, 23,
389-396.
Tayler, B. (2015). Memory Magic: Is it just a bunch of hocus pocus? DECP
Debate, 154, 7-12.
Telzer, E. H., Fuligni, A. J., Lieberman, M. D., & Galvín, A. (2013). The
effects of poor quality sleep on brain function and risk taking in adolescence.
Neuroimage, 71, 275-283.
Thimgan, M. S., Seugnet, L., Turk, J., & Shaw, P. J. (2015). Identification of
genes associated with resilience/vulnerability to sleep deprivation and starvation in
Drosophila. Sleep, 38, 801-814.
Thomason, M. E., Race, E., Burrows, B., Whitfield-Gabrieli, S., Glover, G. H.,
& Gabrieli, J. D. (2009). Development of spatial and verbal working memory
capacity in the human brain. Journal of Cognitive Neuroscience, 21, 316-332.
Thorleifsdottir, B., Bjarnsson, J. K., Benediktsdottir, B., Gislason, T. H., &
Kristbjarnarson, H. (2002). Sleep and sleep habits from childhood to young
adulthood over a 10-year period. Journal of Psychosomatic Research, 53, 529-537.
Tremaine, R. B., Dorrian, J., & Blunden, S. (2010). Subjective and objective
160
sleep in children and adolescents: measurement, age, and gender differences.
Sleep and Biological Rhythms, 8, 229-238.
Unsworth, N. & Spillers, G. J. (2010). Working memory capacity: Attentional
control, secondary memory, or both? A direct test of the dual-component model.
Journal of Memory and Language, 62, 392-406.
Van der Heijden, K. B., de Sonneville, L. M., & Swaab, H. (2013).
Association of eveningness with problem behavior in children: a mediating role of
impaired sleep. Chronobiology International, 30, 919-929.
Van Der Werf, Y. D., Altena, E., Schoonheim, M. M., Sanz-Arigita, E. J., Vis,
J. C., De Rijke, W. et al. (2009). Sleep benefits subsequent hippocampal
functioning. Nature Neuroscience, 12, 122-123.
Vecsey, C. G., Peixoto, L., Choi, J. H. K., Wimmer, M., Jaganath, D.,
Hernandez, P. J. et al. (2012). Genomic analysis of sleep deprivation reveals
translational regulation in the hippocampus. Physiological Genomics, 44, 981-991.
Vedaa, O., West Saxvig, I., Wilhelmsen-Langeland, A., Bjorvatn, B., &
Pallesen, S. (2012). School start time, sleepiness and functioning in Norwegian
adolescents. Scandinavian Journal of Educational Research, 56, 55-67.
Verweij, I. M., Romeijn, N., Smit, D. J., Piantoni, G., Van Someren, E. J., &
Van Der Werf, Y. D. (2014). Sleep deprivation leads to a loss of functional
connectivity in frontal brain regions. BMC Neuroscience, 15, 88.
Vriend, J. L., Davidson, F. D., Corkum.P.V., Rusak, B., McLaughlin, E. N., &
Chambers, C. T. (2012). Sleep quantity and quality in relation to daytime functioning
in children. Children's Health Care, 41, 204-222.
Vriend, J. L., Davidson, F. D., Corkum, P. V., Rusak, B., Chambers, C. T., &
161
McLaughlin, E. N. (2013). Manipulating sleep duration alters emotional functioning
and cognitive performance in children. Journal of Pediatric Psychology, 38, 1058-
1069.
Walker, M. P. & Stickgold, R. (2006). Sleep, memory and plasticity. Annual
Review of Psychology, 57, 139-166.
Walker, M. P. & van der Helm, E. (2009). Overnight therapy? The role of
sleep in emotional processing. Psychological Bulletin, 135, 731-748.
Werner, H., Molinari, L., Guyer, C., & Jenni, O. G. (2008). Agreement rates
between actigraphy, diary, and questionnaire for children's sleep patterns. Archives
of Pediatrics & Adolescent Medicine, 162, 350-358.
Wesnes, K. A., Pincock, C., Richardson, D., Helm, G., & Hails, S. (2003).
Breakfast reduces declines in attention and memory over the morning in
schoolchildren. Appetite, 41, 329-331.
Wiggs, L. & Stores, G. (2004). Sleep patterns and sleep disorders in children
with autistic spectrum disorders: insights using parent report and actigraphy.
Developmental Medicine & Child Neurology, 46, 372-380.
Willgerodt, M. A., Kieckhefer, G. M., Ward, T. M., & Lentz, M. J. (2014).
Feasibility of using actigraphy and motivational-based interviewing to improve sleep
among school-age children and their parents. The Journal of School Nursing, 30,
136-148.
Wolfe, J., Kar, K., Perry, A., Reynolds, C., Gradisar, M., & Short, M. A.
(2014). Single night video-game use leads to sleep loss and attention deficits in
older adolescents. Journal of Adolescence, 37, 1003-1009.
Wolfson, A. R. & Carskadon, M. A. (1998). Sleep Schedules and Daytime
162
Functioning in Adolescents. Child Development, 69, 875-887.
Wolfson, A. R. & Carskadon, M. A. (2003). Understanding adolescents' sleep
patterns and school performance: a critical appraisal. Sleep Medicine Reviews, 7,
491-506.
Wood, M. (2005). Bootstrapped confidence intervals as an approach to
statistical inference. Organizational Research Methods, 8, 454-470.
Worthman, C. M. & Brown, R. A. (2007). Companionable sleep: social
regulation of sleep and cosleeping in Egyptian families. Journal of Family
Psychology, 21, 124-135.
Wu, J. C., Gillin, J. C., Buchsbaum, M. S., Chen, P., Keator, D. B., Wu, N. K.
et al. (2006). Frontal lobe metabolic decreases with sleep deprivation not totally
reversed by recovery sleep. Neuropsychopharmacology, 31, 2783-2792.
Yoo, S. S., Gujar, N., Hu, P., Jolesz, F. A., & Walker, M. P. (2007). The
human emotional brain without sleep: A prefrontal amygdala disconnect. Current
Biology, 17, R877-R878.
163
Appendix 1: Ethics approval UCL RESEARCH ETHICS COMMITTEE GRADUATE SCHOOL OFFICE
Dr Phil Stringer Educational Psychology Group Research Department of Clinical, Educational and Health Psychology 26 Bedford Way UCL 21 May 2013 Dear Dr Stringer Notification of Ethical Approval Project ID: 4519/001: Sleep and working memory in pre-adolescent children I am pleased to confirm that your study has been approved by the UCL Research Ethics Committee for the duration of the project i.e. until August 2016. Approval is subject to the following conditions:
1. You must seek Chair’s approval for proposed amendments to the research for which this approval has been given. Ethical approval is specific to this project and must not be treated as applicable to research of a similar nature. Each research project is reviewed separately and if there are significant changes to the research protocol you should seek confirmation of continued ethical approval by completing the ‘Amendment Approval Request Form’.
The form identified above can be accessed by logging on to the ethics website homepage: http://www.grad.ucl.ac.uk/ethics/ and clicking on the button marked ‘Key Responsibilities of the Researcher Following Approval’. 2. It is your responsibility to report to the Committee any unanticipated problems or adverse
events involving risks to participants or others. Both non-serious and serious adverse events must be reported.
Reporting Non-Serious Adverse Events For non-serious adverse events you will need to inform Helen Dougal, Ethics Committee Administrator ([email protected]), within ten days of an adverse incident occurring and provide a full written report that should include any amendments to the participant information sheet and study protocol. The Chair or Vice-Chair of the Ethics Committee will confirm that the incident is non-serious and report to the Committee at the next meeting.
164
The final view of the Committee will be communicated to you.
Reporting Serious Adverse Events The Ethics Committee should be notified of all serious adverse events via the Ethics Committee Administrator immediately the incident occurs. Where the adverse incident is unexpected and serious, the Chair or Vice-Chair will decide whether the study should be terminated pending the opinion of an independent expert. The adverse event will be considered at the next Committee meeting and a decision will be made on the need to change the information leaflet and/or study protocol. On completion of the research you must submit a brief report (a maximum of two sides of A4) of your findings/concluding comments to the Committee, which includes in particular issues relating to the ethical implications of the research. With best wishes for the research. Yours sincerely Professor John Foreman Chair of the UCL Research Ethics Committee Cc: Rebecca Ashton, Applicant Professor Peter Fonagy, Head of Department UCL Research Ethics Committee, c/o The Graduate School, North Cloisters, Wilkins Building University College London Gower Street London WC1E 6BT Tel: +44 (0)20 7679 7844 Fax: +44 (0)20 7679 7043 [email protected] www.ucl.ac.uk/gradschool
165
Appendix 2: Project date planner
166
Appendix 3: Sleep diary blank form
167
Appendix 4: Actiwatch wearer’s guide
Thank you for agreeing to wear this Actiwatch to track your sleeping habits. You should fasten the watch onto your wrist and just leave it there until you need to return it. Frequently Asked Questions
Can I wear it while I wash?
Yes, the Actiwatch is water resistant so you can wear it in the shower or bath for up to 30 minutes.
Can I wear it while I swim? It’s best to remove the Actiwatch while swimming, as the Actiwatch would be likely to be in the water for more than 30 minutes. Don’t forget to put it back on afterwards!
Can I wear it while I exercise? Yes, the Actiwatch can be worn during any activity where you could wear a normal watch. If your school or club has a policy of no watches during an activity, then you should take off your Actiwatch – but don’t forget to put it back on afterwards!
What are the buttons for? The Actiwatch has some functions which we are not using during this study, so the buttons have been turned off. You can press them if you like, but the buttons won’t do anything.
What is it measuring? The Actiwatch measures light and movement. So it will record when it goes dark and when you stop moving about at night. It will also record any times during the day when you are very still – this is why we need your sleep diary as well, to make sure we only count your sleep time and not when you are just sitting still! The Actiwatch expects some movement every so often, so if you take the Actiwatch off then it will record that you are not wearing it.
Should there be a display on the watch? No, the Actiwatch has been set to have a blank screen for this study.
168
Can I clean the Actiwatch? There should be no need for you to clean the Actiwatch. It will be cleaned by the researcher before being given to you. If you do wish to clean the Actiwatch during the week you have it, you can wipe it with a soft cloth moistened with water and a mild detergent if needed. Do not use bleach, alcohol or abrasive based cleaners as they may damage the Actiwatch.
The display is flashing, what does that mean? The display only flashes if the Actiwatch is not fastened properly. It could be too loose or too tight. Refasten the strap so that the Actiwatch is snug on your wrist.
The display is showing a battery sign, what should I do?
When the battery is running low, this sign will show: Keep wearing the Actiwatch for the rest of your week, and tell the researcher when you give it back.
The Actiwatch is making my skin red, what should I do? First try loosening the wrist strap slightly. If the Actiwatch is still irritating your skin, remove it and give it back. More questions? You can contact the researcher at [email protected] or 01254 666980.
169
Appendix 5: Sleep knowledge questionnaire
Sleep knowledge questionnaire
Please answer the following questions by ticking either true or false
True or False? True False Don’t
Know
1 Sleep helps me remember things T
2 Boys have deeper sleep than girls F
3 Doing something calm before bed can help me fall asleep T
4 I should go to bed at the same time every night T
5 Exercising every day can help me sleep better T
6 I need about 7 hours’ sleep each night F
7 Watching TV before bed can make it hard for people to sleep T
8 Not getting enough sleep can make people overweight T
9 Doing exercise just before bed can help me fall asleep F
10 Children need less sleep than their parents each night F
170
Appendix 6: Amendments to ACES materials Changes are in square brackets, original first then amended wording. Parent booklet p.2 Added information about the UK adaptation after Sarah Blunden’s biography. Contents list added. p.3 “Eat your veggies” replaced with “eat your greens”. p.8 Removed “repeating grades” as we don’t usually do that in the UK. Added “forgetting to take the right kit”. p. 12 Poor sleep in infants - Before [that] [9 months] they actually don’t have a memory of that. p.13 Two additional points in what parents can do to assist children’s sleep:
o Too much light in the bedroom, especially in summer time. Windows should have blinds or curtains that will shut out enough light for the child to sleep even if it isn’t dark outside. Night lights are fine if the child is scared of the dark.
o Being too hot or too cold. It is very hard for anyone to sleep if they are not comfortable. Children need to have ways to change the temperature, e.g. thicker or thinner pyjamas; a sheet if the duvet is too warm; opening or closing a window.
p.13 Changed “gentle loving and touching bedtime routine” to “gentle bedtime routine”. p.16 what parents can do to assist sleep in adolescents – breakfast outside is unusual in the UK!
[Have breakfast outside in the sun.] [Open the curtains or turn the light on when it’s time to wake.] This will give their body the cue to wake up.
p.17 Example – Kelly had to become 18 rather than 16 to reflect UK age for driving. p.18 point 6 – reference to having to repeat a class removed, as this is not a common practice in the UK. p.20 sleep questionnaire – removed item “do you have a bedroom clock?” as most people do and having the clock is not the problem. Changed the next item from “Do you anxiously check the clock when you are awake at night?” to “... check the time” – as people may use other devices than a clock. p.21 getting help for sleep disorders:
4. Allied health care workers (e.g. psychologist) through the [local
171
association] [school nurse or GP] Child workbook p.2 WOW! You are in Year [4] [5]. p.5 Sleep wheel – smaller wheel is labelled “Wheel B – cut this one out” and the instructions relate to Wheel B. p.8 Examples of sleep patterns –gender neutral names have been assigned. The bed times have been amended to match the total sleep times given. The sleepy scores have been changed to match the direction of the scale used (i.e. 2 is quite sleepy and 4 is wide awake). A question has been added at the end of the page: “whose sleep pattern is better?” to clarify the purpose of the information on the page. p.9 Sleep hygiene: [You can include a “special time” with mum and/or dad (reading and cuddling).] [You can include relaxing activities like having a drink, a bedtime story and being tucked into bed.] p.10 Bedroom routine exercise – the example routine has been simplified to make it more realistic for an individual child.
Original example Amended example
Bath or shower- into PJs Bath or shower- into PJs
Warm milk in lounge Warm milk in lounge
Into bedroom Chat quietly with my brother
Spend quiet time Clean teeth
Time with Dad reading Into bedroom
Have a quiet talk to my brother Time with Mum or Dad reading
Listen to quiet music in the dim light Tucked into bed
Turn out the big light – put on the night light
Turn out the big light and put on the night light
Open the door Sleep time
Good night from everyone
Sleep time
p.12 Cloze exercise has been swapped to boxes with only the correct words available. This is to give the best chance of success through errorless learning and makes more readable sentences if the child comes back to the page at a later date. Also removed reference to “ particularly hard things like science or maths” because different children will find different things harder to learn. Also added sentence: your body finds it harder to know how much ………….. (food) it needs, so you can end up overeating and putting on …..……… (weight) p.14 So what have we learned – I have added “3 things I can do to improve my sleep” for children to fill in, following the theory of planned behaviour in which intention to change needs to be operationalised.
172
Just for fun section removed – these quizzes need to be given at set times for evaluation purposes. The relaxation exercise has been moved to just after the bedtime routine exercise. Sleep diary also removed – this will be given as a separate sheet. PowerPoint session 1
Altered page numbers to be consistent with workbooks
5 stages of sleep – added “this is when dreams can happen” to Stage 5 description
Sleep cycle diagram – added “also called Stage 5” to “REM Sleep”
Removed sleep diary slide – children will already have done this before the first session
Placed the “complete the blanks” exercise from www.sleepforkids.org as an extension activity after making the sleep wheel.
Changed final slide from “Australian Sleep Education Programme” to “Australian Centre for Education in Sleep” – just for consistency throughout.
PowerPoint session 2
Changed “student X” and “student Y” to match page 8 of the child workbook
Replaced line graph with bar chart because this is an inappropriate use of a line graph (it doesn’t make sense for the data points to be connected because each point is for a different child):
Replaced by: 0
2
4
6
8
10
12
1 2 3 4 5
Number of students
Ho
urs
of
sle
ep
Sleeping w ell
Not sleeping w ell
173
Children are then invited to identify which children are getting enough sleep – making a more interactive experience for the class.
Altered the “bedtime activities” slide to remove references to their sleep diary, since we are using a simplified diary that does not include activities.
Also on “bedtime activities”, included clearer instructions for teachers about safely asking children to share information about what they did in the hour before bed time last night.
Pie chart removed – I felt it could be confusing for a number of children who may not understand percentages or pie charts well enough yet.
Improving sleep hygiene page – clarified the language: [“Dos “ for good sleep hygiene] [For good sleep hygiene, DO:] and [“Don’ts” for good sleep hygiene:] [For good sleep hygiene, in the hour before bed, DON’T:]
Added a note for the teacher for clarification on the sleep hygiene page: “Doing exercise during the daytime can help you sleep, as you feel physically tired and ready for sleep. However, exercise too close to bedtime can stop you sleeping. So the message is that exercise is a good thing to do, but not right before bedtime.”
Looking at bedtime activities again – I have changed the classroom discussion to focus on generating class ideas for good activities that they could include in a bedtime routine. This brainstorm might help them fill in the workbook page on their own routine.
Own routine is to be done in class rather than at home, to ensure every child completes the activity.
PowerPoint session 3
Improving sleep hygiene slide removed – it’s a repeat from earlier and children can refer to their workbook page.
174
Weblink to sleepsolutions.com removed, internet page no longer exists.
Added suggestion for a Youtube clip on sleep walking. Teacher’s manual and background information
Added information about the UK adaptation after Sarah Blunden’s biography.
Paragraph numbering removed.
Practical guidance placed first, then background information as the appendix (thus reversing the order of the original document).
Practical guidance includes a session plan for each of the 3 lessons (substantially revised from the original, and teachers are encouraged to set out their own lesson plans using their own school’s template during the training session).
Example of a nightmare is provided, to avoid children having to tell their own personal accounts.
Relaxation bathtime exercise removed – it’s in the children’s workbook and doesn’t need duplicating.
Project topic 5 removed, as it refers to adolescent sleep and these are primary children.
Sleep diary and true/false knowledge questionnaires are placed as an appendix, and the rest are deleted as they are only relevant to adults.
175
Appendix 7: Flyer to schools
Getting a good night’s sleep is important for all of us, but especially for children. Sleep helps children to remember things better (Henderson et al, 2012), keep their emotions in check (Simola, 2012) and maintain a healthy weight (Gozal & Kherandish-Gozal, 2012). Crucially for learning, since working memory is a strong predictor of academic performance (Alloway & Alloway, 2010), sleep is related to working memory performance (Steenari et al, 2003). We all know that lots of children are not getting enough sleep and research demonstrates that children are getting less sleep now than in previous decades (Iglowstein, 2003). But what can we do about it? A University College London research study is planned for academic year
2013-14, trialling a sleep education programme, and all Year 5 classes in
BwD are invited to take part! Participating classes will receive everything needed to teach 3 high-quality lessons about sleep and a project to consolidate learning. In return we ask you to help us collect the data required to evaluate this trial programme. For more information please contact: Rebecca Ashton
176
Appendix 8: Teacher information & consent form
Information Sheet for Teachers
This information sheet is for you to keep.
Title of Project: Sleep & working memory in pre-adolescent children
This study has been approved by the UCL Research Ethics Committee (Project ID Number): 4519/001
Name Rebecca Ashton
Work Address
Contact Details
We would like to invite Year 5 classes to participate in this research project.
177
Details of Study:
Getting enough sleep is very important for children, as it affects their learning, health and wellbeing. We are hoping to trial a new sleep programme which will teach children about sleep and how to get into good sleeping habits. The programme involves 3 lessons for the whole class plus a project, with training and all the materials you will need. Schools who volunteer for this programme will be randomly allocated to one of three groups: classroom programme with classroom project, classroom programme with home project and parent information, or waiting list. Schools on the waiting list will be asked to participate in all the data collection, and will get their copy of the intervention programme at the end. The start date for your intervention can be negotiated with me, within the academic year 2013-14. If you wish to participate, we would ask you for some data about each child in the class: their gender, postcode, attainment levels and SEN status. Before and after the intervention, and again 3 months afterwards, each child in the class would do a sleep knowledge questionnaire and would take home and complete a week’s sleep diary. In addition the researcher would visit school during the week before and the week after the intervention, and again 3 months afterwards, to assess each child in the class using computer-based tests of attention and working memory. At the end of the intervention we have a short questionnaire for you, each child and their parents, to tell us what you thought of the programme. We need up to 10 children from each participating class to give us some extra data before and after the intervention, and again 3 months afterwards. This will consist of wearing a motion-sensitive Actiwatch while they sleep each night for a week, to record their sleeping patterns electronically. If you volunteer your class for this study, we would write to each child in the class inviting them to do this extra part. The advantages of participating in this study are that you will receive free training and a free copy of this new sleep intervention programme. We hope, from the existing evidence, that your children will sleep better as a result – but we can’t guarantee that will happen. In return, your commitment will be to deliver the programme in class (and to parents if you are in that group), and to assist us in collecting the data required to evaluate this programme. We will need the use of your ICT suite for a day to do the data collection each time. We don’t anticipate that the programme will cause any problems for your children or their parents, but as with any other curriculum area it may raise issues that you would need to manage, with my support as appropriate.
178
Please discuss the information above with others if you wish or ask us if there is anything that is not clear or if you would like more information. It is up to you to decide whether to take part or not; choosing not to take part will not disadvantage you in any way. If you do decide to take part you are still free to withdraw at any time and without giving a reason. All data will be collected and stored in accordance with the Data Protection Act 1998.
179
Informed Consent Form for Teachers Please complete this form after you have read the Information Sheet.
Title of Project: Sleep & working memory in pre-adolescent children
This study has been approved by the UCL Research Ethics Committee (Project ID Number): 4519/001 Thank you for your interest in taking part in this research. Before you agree to take part, the person organising the research must explain the project to you. If you have any questions arising from the Information Sheet or explanation already given to you, please ask the researcher before you to decide whether to join in. You will be given a copy of this Consent Form to keep and refer to at any time.
Participant’s Statement I (name) ………………………………. (school) ……………………………………
am a Year 5 class teacher.
have read the notes written above and the Information Sheet, and understand what the study involves.
understand that if I decide at any time that I no longer wish to take part in this project, I can notify the researchers involved and withdraw immediately without giving a reason.
consent to the processing of my students’ personal information for the purposes of this study.
understand that such information will be treated as strictly confidential and handled in accordance with the provisions of the Data Protection Act 1998.
understand that participating teachers will receive a copy of a programme which is under copyright, and agree not to share or use it outside this school.
agree that the research project named above has been explained to me to my satisfaction and I agree to take part in this study.
Signed: …………………………………………… Date:
Email: Phone:
I (name of head teacher) ……………………………………………….
agree to support the above teacher and his/her class to participate in this study.
Signed: ………………………………………………………………………… Date:
180
Appendix 9: Parent and child information & consent form
Information Sheet for Parents & Children
You will be given a copy of this information sheet.
Title of Project: Sleep & working memory in pre-adolescent children
This study has been approved by the UCL Research Ethics Committee (Project ID Number): 4659/001
Name Rebecca Ashton
Work Address
Contact Details
We would like to invite Year 5 children to participate in this research project.
Details of Study:
Getting enough sleep is very important for children, as it affects their learning, health and wellbeing. Your Year 5 class is going to try out a new programme of 3 lessons and a project which will teach children about sleep and how to get into good sleeping habits. Everyone in the class will be asked to fill in a sleep diary and a questionnaire as part of this programme. Everyone will also be asked to do computer-based tests of memory and attention skills during school time. These are just part of the curriculum that the school has chosen to provide for Year 5, like any other lessons and like any other homework. However, if you do not wish to take part in this programme please let the class teacher know. We also need up to 10 children from each class to give us some extra data. This will consist of wearing something called an Actiwatch while they sleep each night for a week before the programme starts, a week at the end of the programme and again 3 months later. The Actiwatch looks like a watch, and is worn on the wrist. It measures how long a child is really asleep, as well as any times when they wake up in the night. This helps us to see how accurate the sleep diary is. If you would like to be one of the 10 children who wear the Actiwatch, please return the consent form attached. We will pick 10 at random from the forms we receive, and will tell you if you have been selected or not. The advantages of giving us that extra data are that we will be able to see if the programme makes any difference to children’s sleeping habits, and if getting more sleep means that children are better at paying attention and remembering information. You would be helping us to find out whether this sleep programme really works or not, so teachers can decide whether to use it with their children. In return, you are agreeing to provide that extra data on three occasions. You would also need to give back the Actiwatch at the end of each of the three weeks, as we need them for other children. Please discuss the information above with others if you wish or ask us if there is anything that is not clear or if you would like more information. It is up to you to decide whether to take part or not; choosing not to take part will not disadvantage you in any way. If you do decide to take part you are still free to withdraw at any time and without giving a reason. Whether you take part in this extra data or not, we will write to you at the end of the project to tell you our findings. We should be able to let you know whether this sleep programme improves children’s sleep and, if so, whether that improves their memory and attention.
All data will be collected and stored in accordance with the Data Protection Act 1998.
181
Informed Consent Form for Parents & Children
You only need to return this form if you would like to be one of the ten children in your class who wears an
Actiwatch.
Please complete this form after you have read the Information Sheet and/or listened to an explanation about
the research.
Title of Project: Sleep & working memory in pre-adolescent children
This study has been approved by the UCL Research Ethics Committee (Project ID Number):
4519/001
Thank you for your interest in taking part in this research. Before you agree to take part, the person organising the research must explain the project to you.
If you have any questions arising from the Information Sheet or explanation already given to you, please ask the researcher before you to decide whether to join in. You will be given a copy of this Consent Form to keep and refer to at any time.
Participant’s Statement
I (student) ………………………………………………. (school) …………………………………………………
am a Year 5 student.
understand what the study involves.
would like to be one of the ten children in my class who wears an Actiwatch while sleeping.
understand that I can change my mind about being in this study at any time.
know that my Actiwatch information will not be given to anyone else but the researcher.
Signed: …………………………………………………………………………… Date:
I (name of parent) ……………………………………………….
agree to support the above child to participate in this study.
Signed: Date:
182
Appendix 10: Scattergrams showing relationships between WM and sleep variables
Verbal WM & diary time in bed Visual WM & diary time in bed
Verbal WM & diary time asleep Visual WM & diary time asleep
Verbal WM & diary sleepiness Visual WM & diary sleepiness
183
Verbal WM & actigraphy time in bed Visual WM & actigraphy time in bed
Verbal WM & actigraphy time asleep Visual WM & actigraphy time asleep
Verbal WM & sleep efficiency Visual WM & sleep efficiency